* Implement #33646

* More dox

* Seconds, not milliseconds

* Respect env var values

I broke tests on master with the following change:
0d77aba#diff-205d5528aa8f03902903027631d50ef76459835737fb06b16e55fdd1d7430c8eL331

Relevant failure: https://www.deploysentinel.com/ci/runs/651d290ef6799f2563b8b986

This PR brings it back.

It was deprecated and left in a codebase for backwards compatibility.
But it seems we've migrated all references to the new method `delete-file-if-exists`.

The last time this was touched was three years ago.
Should be safe to remove.

* Fix the flakes for metrics as well

added an example of rounding with a decimal place

`yarn dev` had been adjusted to use `build-hot:js-wait` (which waits for the first CLJS compilation to be done), but `yarn dev-ee` had not been changed, so its JS compilation ran too early, failed, and never re-ran.

We actually had this broken behavior (JS build runs too early, CLJS build completes, JS build fails, JS build notices new inputs and re-runs) for a long time. But since it fixed itself when the JS build re-ran, it didn't cause any issues. Now that the JS build no longer watches the CLJS output for changes, the CLJS output needs to be there for the first JS build.

* fix: add ExpandButton color on parent's hover

* fix: icon's color by canSelect, remove !important

* feat: update expand button style on canSelect props

* fix: icon button color when hovered

* Test using columns from joined tables or cards

* Test multi stage joins

* Test removal on source table change

* Test RHS table column selection

* Reduce flakiness?

* Fix test

* Reduce flakiness?

* Use LeaveConfirmationModal in QueryBuilder

* Add isLocationAllowed prop to LeaveConfirmationModal

* Do not route-block running modified native question

* Fix typo

* Introduce explanatory variables

* Move updating URL during question creation to QueryBuilder

* Schedule navigation to after re-render so that LeaveConfirmationModal prop has a chance to update

* Add a test for unsaved changes warning

* Introduce useCallbackEffect

* Refactor QueryBuilder to use useCallbackEffect

* Refactor ActionCreator to use useCallbackEffect

* Fix useCallbackEffect typing

* Move onSubmit out of the scheduled callback to reduce probability of race conditions

* Add comments explaining useCallbackEffect usage

* Improve JSDocs

* Add a test for running modified question

* Update button query to more accessible one

* Add a test for saving a modified question

* Update test names

* Add a test for leaving with no changes

* Remove isDirty-related no-change assertions, as isDirty flag does not work reliably
- Question.prototype.isDirtyComparedTo would return true because edited "question"
  would have extra "description: null" attribute and no "name" compared to
  "originalQuestion" (see "getIsDirty")

* Move useCallbackEffect to its own directory

* Add test setup for useCallbackEffect

* Add a test case for scheduling

* Add a test for scheduling callback effect

* Do not disable navigation in models
- this is a temporary condition until #33793 is implemented

* Remove redundant _.noop

* Rename isScheduled to isCallbackScheduled

* Replace useBeforeUnload with LeaveConfirmationModal in DatabaseEditApp

* Refactor DatabaseEditApp.unit.spec.js to TypeScript

* Add initialRoute param, return history & add /home route

* Add a test case for showing the unsaved changes modal

* Add a test case for not showing the unsaved changes modal

* Format code

* Extend a test case to assert isDirty being properly computed

* Remove redundant route

* Remove redundant code

* Add a test for custom warning modal when saving edited question as a new one

* Redirect to databases page on database creation in component so that isDirty flag can re-compute on re-render

* Fix mocks setup

* Remove redundant withRouter usage

* Use waitFor + expect + not.toBeInDocument instead waitForElementToBeRemoved

* Add a test case for creating new database connection

* Remove redundant promise

This also moves the CLJS output out of `frontend/src/cljs{,_release}` to
`target/cljs_{dev,release}`. If that causes problems for FE tooling I
don't know about, it could be moved back.

Hot reloading:
CLJS changes are hot-reloaded into the running app by Shadow CLJS - no
action is needed. Note that the UI will not refresh; you usually have to
click around so the updated MLv2 code gets executed.

Usage of the CLJS REPL:
- Open the browser (you should see `shadow-cljs: #N ready!` in console)
- Connect to nrepl on 50655
- "Select" the `:app` build;
    - https://shadow-cljs.github.io/docs/UsersGuide.html#_editor_integration
    - Conjure: `:ConjureShadowSelect app`
    - Plain REPL: Eval in CLJ mode `(shadow/repl :app)`
- Consider running `yarn build-hot:cljs` and `yarn build-hot:js` in
  separate terminals; then when `yarn build-hot:js` crashes you don't
  have to reconnect your REPL.
- The REPL works beautifully, even loading files outside the :entries
  of the shadow-cljs build like tests!

"Stale" error in browser:
- This can happen if the webpack bundle is created before the CLJS is
  done building.
- If you ran `yarn build-hot:cljs` and `:js` separately, let `:cljs`
  finish compiling first.
- If you see this otherwise: (a) contact Braden so it can be fixed; and
  (b) either restart `build-hot:js` or force webpack to rebuild by
  hitting http://localhost:8080/webpack-dev-server/invalidate

Yarn commands:
`dev`, `dev-ee`, `build-hot`, and `build-static-viz:watch` all now wait
for the CLJS build to be complete by watching for files to be generated.
They should work the same from a user's POV.

(Since the webpack build no longer watches the CLJS output,
it won't notice the changes and immediately rebuild after the first
build. This avoids errors on the clean build, or "stale build" messages
in the app from Shadow CLJS.)

Fixes #32674.

* Add `dashboardGrid` E2E helper

* Add test ID for x-ray dashboard header

* Extract 14793 repro to its own file

* Extract 15578 repro to its own file

* Remove duplicate 17968 repro

* Clean up 17968 repro

* Clean up test

* Clean up test

* Clean up test

* Clean up test

* Clean up test

* Clean up test

* Sort imports

* Use shared helpers for join tests

* Extract 12928 repro to its own file

* Minor renames

* Fix `join` e2e helper

* Fix `assertQueryBuilderRowCount` for count > 1000

* Make table check optional in `assertJoinValid`

* Clean up test

* Fix `join` helper

* Fix 17968 test

* Fix 30058 and 32075 tests

* Reduce flakiness?

* Use LeaveConfirmationModal in QueryBuilder

* Add isLocationAllowed prop to LeaveConfirmationModal

* Do not route-block running modified native question

* Fix typo

* Introduce explanatory variables

* Move updating URL during question creation to QueryBuilder

* Schedule navigation to after re-render so that LeaveConfirmationModal prop has a chance to update

* Add a test for unsaved changes warning

* Introduce useCallbackEffect

* Refactor QueryBuilder to use useCallbackEffect

* Refactor ActionCreator to use useCallbackEffect

* Fix useCallbackEffect typing

* Move onSubmit out of the scheduled callback to reduce probability of race conditions

* Add comments explaining useCallbackEffect usage

* Improve JSDocs

* Add a test for running modified question

* Update button query to more accessible one

* Add a test for saving a modified question

* Update test names

* Add a test for leaving with no changes

* Remove isDirty-related no-change assertions, as isDirty flag does not work reliably
- Question.prototype.isDirtyComparedTo would return true because edited "question"
  would have extra "description: null" attribute and no "name" compared to
  "originalQuestion" (see "getIsDirty")

* Move useCallbackEffect to its own directory

* Add test setup for useCallbackEffect

* Add a test case for scheduling

* Add a test for scheduling callback effect

* Do not disable navigation in models
- this is a temporary condition until #33793 is implemented

* Remove redundant _.noop

* Rename isScheduled to isCallbackScheduled

* Remove redundant route

* Remove redundant code

* Add a test for custom warning modal when saving edited question as a new one

The QP supports this - it appears that every type can be `concat`-ed and
it will coerce the values to some kind of string - numbers, dates,
booleans, etc.

* Add `displayInfo` types for cards and tables

* Add TypeScript wrapper for `withDifferentTable`

* Move fields tests under their own `describe`

* Clean up the fields picker

* Add extra test coverage

* Fix missing import

* Fix type

* Migrate `DataStep` to MLv2 (mostly)

* Fix creating a new query

Before
1. prepare-frontend
2. prepare-backend
3. build-static-viz
4. files-changed-check
5. static-viz-files-changed-check

So, we're doing all these expensive operations before the step 4, even though they might not even be necessary.
If files changed already outputs backend_all, this means we have to run these tests whether or not there were static-viz changes.

After
1. test which files changed
    - if backend_all output -> skip static-viz-bundle and start running tests straight away
    - else if there are no direct backend changes, have an additional check for static-viz
2. static-viz-files-changed
    - if true -> run tests
    - else -> skip tests

probably won't be merged. The intent of these changes is to be
explicit in what is passed into each of the card creation functions,
ensure consistency of types, and eliminate redundant or non-required
arguments.

`build-dashcard` was the target of this simplification. The
`available-values` argument is no longer necessary as that data is
sourced from other arguments. the `satisfied-*` arguments are no
longer seqs of satisfied filter, metric, and dimension definitions,
but maps of names of those items to the definitions. This is only
needed because satisfied metrics needs metrics names when adding
metadata. However, all values were modified into a consistent shape.

In other areas, if an aggregate object with several keys is passed
to a function that uses only one element of the object, only the
inner element is passed.

In card-candidates, `:matches` is dissoced from `satisfied-dimensions`
to remove the extra payload that isn't needed in that argument.
This makes it clear that satisfied field values are passed into card
creation functions via the `satisfied-bindings` and not cruft from
earlier stage.

As an aside, one potential fix that would substantially improve the
readibility of this entire codebase would be to move the element
names in the one key to value maps as is done for dimensions,
metrics, filters, and cards into the elements as a named key.
Carrying that shape forward from the yaml files requires a ton of
first this or zipmap that to recover the names into whatever the
current context is.

* Fix overlapping x-tick labels on static viz timeseries charts

* Address review: make the condition easier to understand + fix failed cases

* Update frontend/src/metabase/static-viz/components/XYChart/utils/ticks.ts

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>

* Add a comment to explain why we have to fix timeseries ticks

---------

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>