* Get rid of Question.prototype.aggregate

* Get rid of Question.prototype.breakout

* Get rid of Question.prototype.filter

This removes the `setDefaultQuery()` method for both
- StructuredQuery and
- Query

Resolves #37506

* Migrate segments in getCardForBlankQuestion and... fix it

* Separate parsing, use Metadata.prototype.segment

* Extract filterBySegmentId

* Migrate metrics in getCardForBlankQuestion

* Rename filterByMetricId to aggregateByMetricId

* Expose segment(query, segmentId) in MLv2, map it in metabase-lib, and use it in initializeQB

* Expose metric(query, metricId) in MLv2, map it in metabase-lib, and use it in initializeQB

* Update Lib.metric and Lib.segment return type

* Break TemplateTag type down to ensure presence of snippet-specific and card-specific properties when dealing with respective snippet types

* Begin migrating updateTemplateTagNames, migrate referencedQuestionIds, migrate updateCardTemplateTagNames

* Migrate hasSnippets

* Remove redundant conditioning

* Migrate updateSnippetNames

* Revert "Break TemplateTag type down to ensure presence of snippet-specific and card-specific properties when dealing with respective snippet types"

This reverts commit f303122055ccf7f50ccf7571cec4c5e7a79c5255.

* Migrate updateSnippetsWithIds & setTemplateTag

* Format code

* Use new updateTemplateTagNames implementation

* Remove old updateCardTemplateTagNames, move and refactor relevant test to initializeQB unit test suite

* Remove no-longer-used NativeQuery.prototype.hasSnippets

* Fix TS issues

* Update initializeQB unit tests

* Revert NativeQuery-related changes in initializeQB

* Rename segment to segmentMetadata & metric to metricMetadata

* landing-page backend validation

* fix spacing

* Handle unknown column-settings keys instead of passing a `nil` key.

The issue #27941 called out pivot table static-viz failing when a sort was added to the column-viz.

This led to a null pointer exception in a formatter when the table render was attempted.
Following the chain a bit, I noticed that there's a map of 'db' keys -> 'norm' keys defined in `metabase.shared.models.visualization-settings`

```clojure
(def ^:private db->norm-column-settings-keys
  {:column_title       ::column-title
   :date_style         ::date-style
   :date_separator     ::date-separator
   :date_abbreviate    ::date-abbreviate
   :time_enabled       ::time-enabled
   :time_style         ::time-style
   :number_style       ::number-style
   :currency           ::currency
   :currency_style     ::currency-style
   :currency_in_header ::currency-in-header
   :number_separators  ::number-separators
   :decimals           ::decimals
   :scale              ::scale
   :prefix             ::prefix
   :suffix             ::suffix
   :view_as            ::view-as
   :link_text          ::link-text
   :link_url           ::link-url
   :show_mini_bar      ::show-mini-bar})
```

Which doesn't contain the key `:pivot_table.column_sort_order`, which is the key added by the frontend when you sort
the column via the viz settings.

This would cause the transform to leave a `nil` key in its place which ultimately led to the NPE further down the
line.

I've made the change that any unknown key gets namespaced with `unknown`. So, `:pivot_table.column_sort_order` ->
`:unknown/pivot_table.column_sort_order`.

This prevents the NPE.

* Use a simpler approach to achieve the same result!

* API Keys: add auth card and manage page

* update deletion copy

* add groups data to snapshot

* update e2e tests to work with backend

* update strings and formatting

* update e2e tests

* remove e2e tests from other pr

* update unit tests

* e2e test update

* mock api key endpoints in settingsEditor unit tests

* remove console log

* use dumb quotes

* handle monospace better

* better form validation

* no need to wait

* update test to work with validation

* contrain table width

* remove monospace variant

* API Keys: Handle Groups UI (#37099)

* API Keys: Show in Group members table

* API Keys: Show count in Group listing

* tweak and test group settings views for api keys

* fix sad lint rebase sitch

* update group manager tests

* better loading behavior

* handle all users api keys

* update types and tests

---------

Co-authored-by: Ryan Laurie <iethree@gmail.com>

---------

Co-authored-by: Ryan Laurie <iethree@gmail.com>

As a follow on to #37546, this fixes an issue in which the macroexpansion of `with-skip-if-empty-pulse-result` doesn't capture the scope of the arguments in `skip-if-empty-test`.

The solution is just to _correctly_ wrap all of the `mt/id` functions in `mt/dataset`.

* fix color button when it's disabled (#37053)

* new utm tags for embed settings (#37403)

* new utm tags for embed settings

* use paid plan utm sourcein premium unit tests

* fix e2e

* revert un-needed changes

* Use new docs url for quickstart link (#37523)

* use new docs url

* use realistic version numbers instead of latest for doc urls

* Migrate `isEmpty` check for structured queries in DatasetEditor

Resolves #37515

* Remove `isEmpty` method from StructuredQuery

* Remove related unit test

Fixes #37461.

Resolves #37555

* Migrate EmbedModalContent to TS

* Add basic UI changes to Embed modal

* More styling

* Add more styling

* Self review fixes

* Self review fixes

* Fix action button

* Fix unit tests

* Fix e2e tests

* Fix e2e tests

* Code review and design fixes

* Fix unit test

* Fix for e2e tests

* Code review fixes for e2e tests

* Fix for e2e tests

Resolves #37453

* Validate json settings during init (#36182)

* Use correct translation for error message

Co-authored-by: Cal Herries <39073188+calherries@users.noreply.github.com>

* Make extensible, only throw on env vars

* Better check for whether a setting came from an environment variable

Co-authored-by: Cal Herries <39073188+calherries@users.noreply.github.com>

* Resuse existing parser validation, improve security

* Add example of better exception

* Add precise exception messages as exmaples

* DRY up tests

* Add test for realize

* make the example actually deep

* better hygiene around ex-cause

* Use strict json parsing for all settings

* Disable validation on non-json configuration

---------

Co-authored-by: Cal Herries <39073188+calherries@users.noreply.github.com>

* Add TS wrapper for CLJS `native-query`

* Add TS wrapper for CLJS `raw-native-query`

* Add TS wrapper for CLJS `with-native-query`

* Add TS wrapper for CLJS `with-template-tags`

* Add TS wrapper for CLJS `template-tags`

* Add TS wrapper for CLJS `extract-template-tags`

* Add TS wrapper for CLJS `has-write-permission`

* Add TS wrapper for CLJS `with-different-database`

* Add TS wrapper for CLJS `engine`

* Fix types

* Replace the direct ML reference with a new wrapper

* Simplify the return type for `extractTemplateTags`

* Don't send results works at the card level

When a user enables **Don't send if there aren't results** this was the previous behavior:
- If no results appeared on any card, no emails was sent. This was expected.
- If any card contained results, all cards were sent in the email, including those that had no results.

This PR modifieds that behavior to only send those cards that contain results for a dashboard pulse. For example, if 5 cards are present on a dashboard and 3 have no results, only results for the 2 cards containing data are included in the pulse.

Primary changes:
- `metabase.pulse/execute-dashboard` now removes any card with a row count of 0 in its computed result if `skip_if_empty` is true for the pulse.
- The addition of tests covering the matrix of tests where `skip_if_empty` is true or false, all cards contain data, some cards contain data, and no cards contain data.
- The `with-skip-if-empty-pulse-result` makes the above set of tests nice and concise.

Minor changes:
- Fixing a minor issue in `with-metadata-data-cards` in which the macroexpansion logic would go bad if the user used a different symbol name for `base-card-id`.
- Spelling and formatting issues in `metabase.pulse`

Fixes #34777

* lint fix for #_{:clj-kondo/ignore [:unresolved-symbol]}

* rough approach for per-namespace schema checking

* Added tests

* cleanup + add a namespace to skip

* use stringified namespaces instead of ns objects

* move to `instrument-ns?`+ trying to not emit an if

* WIP, working on tests

* eval

* Fixup tests and remove some errant junk

tests are a bit tricky because the the reader expands the macro so we
can't do anything at runtime to affect the binding. This is why we use
eval so we can coordinate when the binding happens

* some notes and remove the pulse ns from ad-hoc exclusions

* add docstrings and comments

* remove redundant case statement

* swapped branches in mu/defn, so that when `skip?` we emit deparameterized fn

* skip logging message about calculating display name in prod

- it was suspected as part of the high cpu usage issue here:
  https://github.com/metabase/metabase/issues/37144



* config/is-prod? is not avaliable in cljs

* use the right kind of quote

* actually it's ok to use binding if we delay the evaluation of the macro

* use :cljs-dev and :cljs-release to skip printing sus message in prod

* use cljs-dev and cljs-release to control printing message

* add more tests for defn, and move fn_skip_test into fn_test

* add info in mu/defn docstring about instrument-ns? skipping behavior

* fix test

---------

Co-authored-by: dan sutton <dan@dpsutton.com>

* updates enterprise appearance settings ui to validate that landing page urls are relative

* adds e2e test, impls feedback to use isMetabaseUrl util

* makes new e2e tests more resilient

* fixes logical mistake in e2e test

* removes old e2e test that accounted for invalid landing page urls being saved

* impls pr feedback

* extracts out url validation logic and adds unit tests