* Reinforce various `question` tests that were failing in GHA

* Add `question` E2E` group to the GitHub Actions on `master`

* update timeout and use buildjet

* return percy push job branch constraints

* revert timeouts

* use prepare-cypress

* Upgrade `@percy/cypress` to the latest version `3.1.1`

* Upgrade `@percy/cli` to the latest version

* Adding currents

* Update .circleci/config.yml

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

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

* Add validation logic to TemplateTagDimension dimension tag handling

* Add basic error message to query builder

* Update error text

* Add unique error message for validation errors

* Update text of error state

Co-authored-by: Maz Ameli <maz@metabase.com>

* Fix types

* Fix NativeQuery unit test

* Add unit tests for new components

Co-authored-by: Maz Ameli <maz@metabase.com>

* Fix sandboxing test related to subscriptions

* Fix subscriptions test related to joins

* Fix repro 12457 flake

* Fix repro 17397 flake

* Remove redundant tests

* Fix a regression from fixing #20228

* Reorder code order to make it easier to reason about.

* Containerizing using the current release container process, as it's not only more representative, but also we use yarn and maven caching which should be faster

* rebasing

* Update uberjar.yml

* docs - update sql editor page

* change sql example to upper case

* Apply suggestions from code review

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

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

* Add tippy/popper replacement for PopoverWithTrigger

* Refactor and fix types

* Move PopoverWithTrigger components to own folder

* Add tests

* rename children prop to popoverContent

* Fix direct ParameterTargetWidget.jsx reference

* Add a guard that reduces flakiness of binning tests

* Increase timeout for QA DB sync

* Delete DB after the snapshot was taken

This is a hack that seems to prevent race conditions when one DB's sync interferes with the subsequent DB's sync.

* Get slack users and conversations simulatenously

- instead of waiting for the channels, and then getting the users

* Add cache+timeout capability to conversations-list

* extend async approach to users-list

- return timeout+cache warning
- adds tests

* updating docstrings + adding more tests

* Add cache clear and seed slack token settings on-change

- add dox for :on-change for defsettings
- fixing tests

* Fixing tests

- cannot pass a declared (not def'd) value
- some cleanup

* remove unused function

* remove need for slack-configured from timeout fns

- update some docstrings

* pre-populate slack caches

- on app init

* introduce blocking (and cache resetting) endpoint

- this one does not timeout, meant to be called when a user hits a
  button like: "Refetch my slack channels"

* removes -timeout versions of users/channels fn

- the endpoint now returns the cached slack data
- removes tests

* fixes routing layer

* remove trailing whitespace

* add missing docstring

* move the cached slack data into a setting

* add missing docstring

* fix typo

* update form-input-test

* refresh cache in test

* remove un-needed anonymous fn wrapping

* refresh slack cache when user sets new valid token

* refresh the cache when reading from it

- the more its used the more up to date it will be

* code review responses + removes 2-arity refresh-cache!

- clear out cache when slack connection is broken or unset
- add to docstring for slack settings route
- wrap init refresh-cache! with a future
- re-spell parallel

* do not block when returning cached slack data

* rename cache defsetting

* Addressing code review issues

- Adds scheduled updates to the slack users and channels cache
  every 4 hours
- Removes `form_input_refresh` route
- Automatically refreshes cache when `form_input` route is hit,
  if 10 minutes have passed since the last cache fix
- Removes cache refresh on app startup
- Stores only user names, and channel names in the cache, not maps with
  irrelevant data
- Clears out some noisy linter warnings via `#_:clj-kondo/ignore`

* Adds slack task RefreshCache jobs

- Sets up startup-triggered and cron-triggered (every 4 hours at a
  random minute) slack channel/username cache refreshing

* Implements 2 jobs to update slack users/channels

- one on startup, one every 4 hours
- improve docstrings

* ns linter fixes

* more ns linter fixes

* refactor files-channel to look in the cache

- instead of making a tricky api call
- fixing tests

* update files-channel-test to test the cache

* fixup integrations.slack ns decl

* fixup test integrations.slack ns decl

* fix outdated mock of slack/files-channel

* fix a few more tests

* fix api slack-test

* adds locking to refresh-channels-and-usernames!

* re-order requires

* Use Service Accounts for Google Analytics auth going forward

* Update GA dox

* Dox tweak

* Fix e2e test

* Remove dead googleanalytics form code

Co-authored-by: Dalton Johnson <daltojohnso@users.noreply.github.com>

* Fix `[:relative-datetime :current]` inside `:between` filter clause

* Update shared/test/metabase/mbql/util_test.cljc

* Update shared/test/metabase/mbql/util_test.cljc

* download permission

* update specs

* remove console log

* fix types

* lint