Supporting "seconds", "minutes", and "hours" for relative date strings under "past", "next", "last", and "this" in the backend

Add the "minutes" and "hours" options to the frontend widget for relative date picker

Adding backend parsing tests for new options

Adding Cypress test to ensure new options appear in the relative date filter widget

#12492 Repro: Custom expression filter with dates wrong display on subsequent click [ci skip] (#14772)

Fixes #14724 × FK remaps
Fixes #14724 × inception query
Fixes #14724 × inception query × FK remaps
Fixes #14766

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

* "regex-match-first" and "replace" operators aren't safe against SQL injection

DRIVER SPECIFIC CHANGES:
Updating Redshift driver code to use parameterized search/replace patterns for :replace, and the result of running the QUOTE_LITERAL function on the raw string for :regex-match-first

Updating Redshift driver version to latest

Parameterizing pattern in Postgres version of :regex-match-first implementation

TESTS:
Adding new tests to string_extracts_test.clj for ensuring :replace and :regex-match-first work with single quotes in expressions, across all DB drivers supporting those features

Adding ability to pass filter into test-string-extract helper, using that from new tests

Fixing typo in name of test-replace

* Fixing import order in namespace declaration

Adding explicit group number param of 0 for Hive-like driver (since the default is 1), when calling regexp_extract

* Remove trace logging in the new test

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

Changing the namespace name of the logging middleware broke access
logging. Changing this config file to the proper namespace for logging
makes it happen again

See #14773

This field is always plain text to let other instances of mb know they
have to flush the settings cache.

#14766 Repro: Cannot summarize columns from joined table based on a Saved Question [ci skip] (#14770)

* Fix metadata for native sandboxes

* Relax the restrictions

* Another fix 



* Revert accidental commit

* Update repro for #14612

Throw an error for differently-typed columns.

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

* use lighten() for PIVOT_BG_DARK and PIVOT_BG_LIGHT

* apply TableInteractive-cellWrapper class to pivot table cells

* better import

* Run repro for #14724 in 2 versions: "remapped" and "default"

* Use helper function for value remapping

* Export `remapDisplayValueToFK` function

* Pass `name` field as an argument in a function

* Allow for `quick` parameter on `notify/db/:id` for quick syncing

* Ensure auth-ed in api tests

* Move to mt/client

* Full sync / metadata sync of db

* Update tests

* errant comment no longer applicable

* Rename parameter to {:scan full} or {:scan schema}

* Update api/notify/db docstring and remove extra require

* Use defendpoint schema enforcement instead of bad custom one

* Run tests synchronously

* Remove custom macro for `mt/with-temporary-setting-values`

* Ensure request has non-nil api-key before checking

* Incorporate feedback from cam about macros

- it is super important that the db/table-syncs are off thread and the
request is not blocked. Getting that macro wrong would be devastating
and it only exists so it can be written `(execute (db-sync-fn
database))` instead of *gasp* `(execute #(db-sync-fn database))`. Used
in three places, and getting it wrong is terrible. Delaying into the
thunk makes clear that this will work just fine.

* Quick sync

* Try reducing -Xmx for :ci profile a bit to prevent random test failures

* Reenable test

* Query results metadata should include field refs (Fixes #10474)

* Test fixes 

* Test fix 

* Try unskipping some EE tests, let's see what gets fixed

* Try lowering heap size limits on CI a bit to see if things will actually run.

* Test fix

* Re-skip 13641 and 13642 tests

* Make sure resolve-joined-fields works on all levels (fixes #13642)

* Fix missing :require

* Remove logging

* Re-enable test

* Add ->$ids and ->query-shorthand test util macros

* Fix 13641

* Text fix

* Revert changes to backend.js

* Test fixes 

* Remove stray printlns

* Test fix 

* Fix saving a Card, add test

* Fix saving a Card, add test

* Test fix for non-H2 drivers

* Test fixes 

* Revert accidental commit of experimental test macros

* Fix occasional test failure 

* Fix occasional test failures with metabase.api.dataset-test/basic-test

* Fix occasional test failure 

* Another driver test fix 

* Re-enable test for #14724

* Extract remap logic into a separate function

* Run repro for #13642 in 2 versions: "remapped" and "default"

* Update repro for #14724

* Run repro for #13641 in 2 versions: "remapped" and "default"

* Fix 13642 with FK remaps in place

* Test fix 



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

* Quick sync

* Try reducing -Xmx for :ci profile a bit to prevent random test failures

* Reenable test

* Query results metadata should include field refs (Fixes #10474)

* Test fixes 

* Test fix 

* Try unskipping some EE tests, let's see what gets fixed

* Try lowering heap size limits on CI a bit to see if things will actually run.

* Test fix

* Re-skip 13641 and 13642 tests

* Make sure resolve-joined-fields works on all levels (fixes #13642)

* Fix missing :require

* Remove logging

* Text fix

* Revert changes to backend.js

* Test fixes 

* Remove stray printlns

* Test fix 

* Fix saving a Card, add test

* Fix saving a Card, add test

* Test fix for non-H2 drivers

* Fix occasional test failures with metabase.api.dataset-test/basic-test

* Fix occasional test failure 

* Another driver test fix 

* Extract remap logic into a separate function

* Run repro for #13642 in 2 versions: "remapped" and "default"

* Fix 13642 with FK remaps in place

* Test fix 



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

* Parse mb-db-connection-uri when it contains a password

Using the raw jdbc string when its of the form
`username:password@host:port` causes problems if passed directly to
jdbc. Clojure.java.jdbc has some special handling for username and
passwords

```clojure
(if-let [user-info (.getUserInfo uri)]
  {:user (first (str/split user-info #":"))
   :password (second (str/split user-info #":"))})
```

Our logic here is:
1. if mb-db-connection-uri is set and not an older style password,
return that, possibly prepending jdbc: to it
2. if mb-db-connection-uri is set and is an older style passowrd, then
we use the old parsing code which doesn't support all of the crazy
stuff individual connection strings may offer and return a db-spec
3. if all the individual parts are supplied (mb-db-host, mb-db-port,
etc) are used to construct a db-spec

* Correct what old style means for conn-uri

accidentally thought user@host/metabase?password=password was ok. Any
credentially in the host spot is not supported

* Cleanup documentation

don't refer to old password style but call it inline credentials

* Fixups for connection parsing

- ensure we turn postgres: to postgresql: in connection string
- ensure we warn about postgres ssl edge case when using raw
  connection string
- make warnings testable

* Space in inline credentials warning

* Strip jdbc from connection when checking if inline credentials

* Fix for old credential style

* Use log instead of printf/println for copy

* Use log/warn for the message about H2 database already existing

* Remove compare-h2-dbs -main entrypoint

* Clean namespace decl

* Feedback from @walterl

* Fix occasional test failures with metabase.api.dataset-test/basic-test

* Fix occasional test failure 

* Quick sync

* Try reducing -Xmx for :ci profile a bit to prevent random test failures

* Reenable test

* Query results metadata should include field refs (Fixes #10474)

* Test fixes 

* Test fix 

* Try unskipping some EE tests, let's see what gets fixed

* Try lowering heap size limits on CI a bit to see if things will actually run.

* Test fix

* Re-skip 13641 and 13642 tests

* Revert changes to backend.js

* Fix saving a Card, add test

#14726 Repro: Custom filter expression does not show original editor on subsequent click [ci skip] (#14728)

#14649 Repro: Cannot add subsequent aggregation on a custom expression in notebook (#14723) [ci skip]

* Sort GeoJSON maps

* Fixup syntax, removing $FlowFixMe

not sure if this is a no-no but `yarn lint-flow` succeeded for me

* Use pairs and watch for null names

* Quick sync

* Try reducing -Xmx for :ci profile a bit to prevent random test failures

* Update quick sync parameter name

* Update docstring for dataset

* Fix sync-database! schema

* Fix schema again

* Test fix 

* Rework tests in api.gtap-test

* Return actual error message when trying to save an invalid sandbox query

* Minor refactor

* GTAP modal should display errors

* Fix flow failure

* Test fix

* Another flow fix 



* Longer timeout for the test that keeps failing

* Return error message as :message

* New cache key for drivers/uberjar so Oracle/Vertica errors stop popping up

* Revert cache key rotation

* #14612 Repro: Sandboxing limitations should have meaningful UI error message/feedback (#14690)

* Fix random Cypress test failures when fingerprinting fails?

* Bump no-output-timeouts to 15m because CircleCI is extra slow lately

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