* Fix custom field type aggregation not showing

When the origin table don't have that field type.

* Add tests

* Remove puzzle driven development comments

* Make `getAggregationOperators` API more apparent

Co-authored-by: Mahatthana (Kelvin) Nomsawadi <me@bboykelvin.dev>

Co-authored-by: Emmad Usmani <emmadusmani@berkeley.edu>

* Transform card data into the same format as that which comes out of the DB

* Use `from-json` instead of toucan transformations

* Remove unused require

* Update docstring and rename to from-parsed-json

* Oops, fixed

* Remove now

* Add docstring

* Fix order of docstring

* Don't update created_at and updated_at

Co-authored-by: Aleksandr Lesnenko <alxnddr@users.noreply.github.com>

Co-authored-by: Ngoc Khuat <qn.khuat@gmail.com>

* Extract `parseArguments` into utils

* Allow custom browser for Cypress snapshots generation

Co-authored-by: Nemanja Glumac <31325167+nemanjaglumac@users.noreply.github.com>

Co-authored-by: Nick Fitzpatrick <nick@metabase.com>

* Updating graph.metrics when column is removed

* updating logic and adding unit tests

Co-authored-by: Nick Fitzpatrick <nick@metabase.com>

* Ensure we use the ssh tunnel on action execution

I'm gonna see if we have an existing way to run tests against ssh dbs. I
see there are some tests that use a faking ssh server. Not sure if those
are useful or not. But otherwise ssh stuff needs out of process help to
accomplish so not sure if it's easily feasible right now

* introduce new `with-unpooled-connection-spec` macro

it is identical to the `with-connection-spec-for-testing-connection`
macro but has a better name. In the future they could diverge as testing
requirements and action requirements change. But we want that
_implementation_ (which correctly uses tunnels), but we do not want a
"testing connection" slug in the usage.

In the future (or now if we prefer a different name) we can rename it,
find the usages that are not testing vs the ones that are, etc.

* tests for actions over ssh

* unused require of mb.u.ssh

had brought it in but then stuffed all of this inside of
`with-unpooled-connection-spec` so no need to worry about ssh, closing
it, ensuring the port is there, etc.

* Don't test h2 with ssh tunnel

details lack a host and it npe's in the ssh tunnel stuff. No big loss as
ssh on h2 is not supported (due to this error, and probably many more)

* Switch to toucan2 for db access in tests

* Use connection pool for custom actions connection

Was originally using `jdbc/get-connection` from the details of the
db. This was a one-off non-pooled connection that also did not reuse ssh
tunnels. I thought the one-off connection was important since it is not
read only, but that's not the case. So we can reuse the exact same
mechanism that the implicit actions are using with
`metabase.driver.sql-jdbc.actions/with-jdbc-transaction`.

* cyclic dependency

almost moved it into execute, but that gives a very inviting
`sql-jdbc.execute/with-jdbc-transaction`. Welcoming even. But
`sql-jdbc.actions/with-jdbc-transaction` lets you know its only for
transactions.

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

Co-authored-by: Jeff Bruemmer <jeff.bruemmer@gmail.com>

Co-authored-by: Nemanja Glumac <31325167+nemanjaglumac@users.noreply.github.com>

Co-authored-by: Natalie <nat@metabase.com>

* Fix empty database picker state

* Remove redundant `requireWriteback` occurrences

* Update test

* Update test

* Use clearer names

* Don't show query language message for actions

* Use database `hasActionsEnabled` method

* Remove `requireWriteback` prop

* Respect `readOnly` prop in `DataSourceSelectors`

* Extend e2e test coverage

* Use `setActionsEnabledForDB` in Cypress

* Fix data picker behavior

Co-authored-by: Anton Kulyk <kuliks.anton@gmail.com>

Co-authored-by: Nemanja Glumac <31325167+nemanjaglumac@users.noreply.github.com>

* remove 'category' as a field type, because it isnt

* update types

Co-authored-by: Ryan Laurie <30528226+iethree@users.noreply.github.com>

Co-authored-by: Alexander Polyankin <alexander.polyankin@metabase.com>

Co-authored-by: Alexander Polyankin <alexander.polyankin@metabase.com>

Co-authored-by: Jeff Bruemmer <jeff.bruemmer@gmail.com>

Co-authored-by: Ryan Laurie <30528226+iethree@users.noreply.github.com>

* Handle data permission error for dashboard actions

* Extract `setupUnauthorizedDatabaseEndpoints`

* Type error as `unknown`

Co-authored-by: Anton Kulyk <kuliks.anton@gmail.com>

Co-authored-by: Bryan Maass <bryan.maass@gmail.com>

* adding hasMultipleDimensions check to hasLegend clause

* Adding queryBuilderMain helper

* PR Feedback

* remove only

Co-authored-by: Nick Fitzpatrick <nick@metabase.com>

* fix exclude date filter serialization

* review

Co-authored-by: Aleksandr Lesnenko <alxnddr@users.noreply.github.com>

* Allow linkviz to show recent item suggestions

* unit test recent link adding

* oops

Co-authored-by: Ryan Laurie <30528226+iethree@users.noreply.github.com>

* Users can execute actions if they can read the model and the DB is not blocked

[Fixes #28425]

Co-authored-by: Tim Macdonald <tim@metabase.com>

* Optimize average case DB calls

* Typo

* Fix test

* Remove duplicate require

* Return nil

* Fixes

* Fix docstring

* Remove unused require

* Avoid hydrating ordered_cards on the dashboard again

* Tidy

* Update docstring

* Move transaction

* Move comment

* Remove dashboard->param-field-ids

* whitespace

* Tidy update-dashcards!

* Updates from suggestions

* Update test

* Update test description



---------

Co-authored-by: Ngoc Khuat <qn.khuat@gmail.com>

* Use icon helpers

* Remove redundant export

* Use top-level component in tests, keep clean state

* Remove redundant `findBy*` lookups

* Extract IDs to constants

* Reduce repetitive code

* Always mock execution endpoint

* Get back changes lost while rebasing

* Don't repeat exec mock URL everywhere

* Remove retrieve-dashboard-card call

* Have `update-dashboard-card!` return nil, not true

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

* try fixing collections flake

* update spec

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

* Add migration for deleting the abandonment email task

* Start+stop scheduler around the migration

* Update clj-kondo

* Rename migration

* Change to 46

* Fix define-migration

* Add test

* Update ns description

* Remove stop-scheduler! from migration

* Remove migration dependency on metabase.task

* Restore mdb.connection -> mdb

* Update custom-migrations ns docstring

* Remove unused code

* Fix comment

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

Co-authored-by: Gustavo Saiani <gus@metabase.com>

Co-authored-by: Alexander Polyankin <alexander.polyankin@metabase.com>

Co-authored-by: Jeff Bruemmer <jeff.bruemmer@gmail.com>

Co-authored-by: Natalie <nat@metabase.com>

* Fix flaky test and test for a new case

flaky test was annoying. It would timeout after 45 seconds. And
annoyingly, it times out on the _client_, not the server. So we don't
get the proper error message we expect.

And what's happening is that the backend just hooks up the pipe from
it's request to the response. And if no bytes ever come across, we just
twiddle our thumbs until the http request in the test gives up.

The only reliable way i could think of to fix this is to spin up a
webserver that just accepts and then ignores requests (which is what
github did to us in the flaky test).

Had to bump the ring adapter to get the newly exposed `:async-timeout
60000` parameter. By default the webserver in the test has a timeout of
30 seconds, which is shorter than the 45 second timeout for the test's
request to the geojson endpoint.

* add comment

* clean up test server logic

---------

Co-authored-by: dpsutton <dan@dpsutton.com>
Co-authored-by: Nemanja Glumac <31325167+nemanjaglumac@users.noreply.github.com>

* only save changed dashboards

* test early return from save action

* remove problematic api wait for dashboard save

* deeply compare cards and add better tests

* update failing e2e test

Co-authored-by: Ryan Laurie <30528226+iethree@users.noreply.github.com>

Fixes #25378

When using a relative filter with a starting from n units ago on an
aggregation with the same units, we were incorrectly assuming that we
would be working with a field ref at the top level. However, in this
scenario the field is nested such as `[:+ [:field $created_at
{:temporal-unit :month}] [:interval 1 :month]]`. This change makes sure
that we are altering the field-ref.

Co-authored-by: Case Nelson <case@metabase.com>