Merge x.39.3 -> master

* Fix question archiving test

* Fix archiving a question

* Add repro note for #16512

* Add repro for #16496 (bulk archive broken)

* Add E2E test for bulk move

* Fix bulk archive

* Remove reduntant checks

* Use `cy.button()`

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

* Add repro for #16497 (pinned items bulk selection)

* Fix pinned items bulk selection

* Remove redundant check

* Refactor existing "bulk actions" tests

* Extract test that is not related to bulk actions

* Add `pinAllRootItems()` helper function

* Unify "bulk selection" tests

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

* Add helpers to define Settings with multiple impls

* Add docstrings

* Fix missing :require

* Add namespace docstring per @dpsutton suggestion

Friendly names bugfix 1 was a noop because of a typo in the WHERE clause of the DB migration. This one has been manually tested to go and sync on all 4 supported MB db's.

* changes to sync-group-memberships! and unit tests

* update jwt and saml code

* fix admin sync tests

* fix misplaced docstrings

* address review comment

Repro:
db setup:

```sql
create table wut2 (t bytea, ts text);
insert into wut2 (t, ts) values ('20210421164300'::bytea, '20210421164300');
```

Mark t as :Coercion/YYYYMMDDHHMMSSBytes->Temporal, ts as
:Coercion/YYYYMMDDHHMMSSString->Temporal

Do a simple query, and then aggregate count by the two dates. ts
should succeed and t fails.

We are passing in the UTF8 as a parameter rather than just emitting a
literal.

Before:

```clojure
{:sql
 "-- Metabase\nSELECT CAST(to_timestamp(convert_from(\"public\".\"wut\".\"t\", ?), 'YYYYMMDDHH24MISS') AS date) AS \"t\", count(*) AS \"count\" FROM \"public\".\"wut\" GROUP BY CAST(to_timestamp(convert_from(\"public\".\"wut\".\"t\", ?), 'YYYYMMDDHH24MISS') AS date) ORDER BY CAST(to_timestamp(convert_from(\"public\".\"wut\".\"t\", ?), 'YYYYMMDDHH24MISS') AS date) ASC",
 :params ("UTF8" "UTF8" "UTF8"),
 :type :invalid-query}
```

After:

```sql
SELECT CAST(to_timestamp(convert_from(\"public\".\"wut\".\"t\", 'UTF8'), 'YYYYMMDDHH24MISS') AS date) AS \"t\",
       count(*) AS \"count\" FROM \"public\".\"wut\"
GROUP BY CAST(to_timestamp(convert_from(\"public\".\"wut\".\"t\", 'UTF8'), 'YYYYMMDDHH24MISS') AS date)
ORDER BY CAST(to_timestamp(convert_from(\"public\".\"wut\".\"t\", 'UTF8'), 'YYYYMMDDHH24MISS') AS date) ASC
```

Note there's a lot of extra casting that perhaps we can clean up

Previously just used JDBC. However, JDBC will merrily execute queries on empty file locations for SQLite, because it'll create a new SQLite file. This one checks that there exists a SQLite file already to be querying in the first place

Folks wanted DB pagination in #11806 and apparently others but it didn't make it to notion. Well, it made it to my attention, anyhow, and here is some DB pagination, but just for the backend.

* adjustments to CSP header in dev environment to make react dev tools work on Firefox

* add test for script-src directive

* fix position of doc string

* fix error in common_test

There was no DB migration for previous friendly field name defriendlification. This adds DB migration.

Signify this is for internal use for the Metabase team, and remove ETA from the template (since that's tracked outside of GitHub).

#16350 Repro: "Invite someone" button overlaps activate/deactivate radio buttons on small screen widths (#16355)

Peeps want to make questions faster, but we don't give the data about query execution in a very user-friendly way. Here's it in a user-friendly way.

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

* Add the last-edit-info to the child query

just for cards right now, need to extend to dashboards and then make
sure we clean it up into the proper map.

Have a set of all columns that are required and ensure those are in
each query. Nulls need to be cast to the correct type if they aren't a
text column (at least on postgres, need to investigate h2 and mysql).

* Correct find last-edit-info for cards and dashboards

* Move from group-by max id to self join

* Paginate and search on collection items

- include the metadata information in the query
  - to accomplish this, make a backstop to add all columns into each
    query
  - need types to union the null. in postgres it assumes all nulls are
    text. Haven't checked what mysql and h2 do here yet
- sort_column: name, last_edited, model
- sort_direction: asc, desc

* Add default sort column and direction

* handle mysql

mysql doesn't need types on nulls to unify them

* Correct batch post processing for cards and dashboards

* Correct order of expectations

they were alphabetical inside of models
previous order:
- acme products           | pulse
- electro-magnetic pulse  | pulse
- birthday card           | card
- dine & Dashboard        | dashboard

and is now ordered solely by name

* Tests and sort on lower name after timestamp

When sorting by last edited, lots of things don't have this, so don't
want arbitrary sorting, but sort by name afterwards

* Ensure nulls go to the end of the list

* Tests

* trying to make h2, mysql, and pg happy

sorting timestamps and always having nulls last is quite difficult

* Make diff bigger

this was going in as a string and losing the timezone information, so
was actually advancing time.

Now: (java.time.OffsetDateTime/now)
"2021-06-01T13:59:33.165483Z"

Minus 2 hours: (.minusHours (java.time.OffsetDateTime/now) 2)
"2021-06-01T11:59:27.528500Z"

As seen in the code:
;; unaltered "now"
"2021-06-01T08:55:28"
;; altered "two hours ago"
"2021-06-01T11:55:29" <- UTC shenanigans. so just bump back 24 hours

* Sort by model ranking

* Remove clj-time and use ZonedDateTime instead of OffsetDateTime

...since the back-end is already capable of handling those.

* Handle PK/FK of the type STRING or STRING_LIKE

* Warn users about sandboxing and public sharing

* Update docs/enterprise-guide/data-sandboxes.md

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

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

* Add type on collection

* Search with collection type

* Cleanup bounded heap accumulator

* Make search conditionally bump official collection items

* collection api and tests

* Put collection type onto hydrated cards on dashboards

* Move official collection type scoring into ee codebase

* ensure ee and oss agree when not official collection

* Mark Collections tree with type

* Remove unnecessary `and`s when no check for enhancements

* Tests for setting collection tree type

* Include hydration for collection type on dashboards

* Make sure created test collections are cleaned up

* Cleanup tests, don't search on collection_type

looks for all text fields and searches the term against that. official
would bring back all official types. no bueno

* Docstring on protocol and don't shadow comparator

* update to new ee impl var passing style

* Collection model ensures no type change on personal collection

* Check for personal collection when updating type

model checks for personal collection and rejects in the update but
doesn't check for children of personal collection.

* Update checking for unchangeable properties of a personal collection

* Cleanup: type collection tree, batched hydration, combine error checks

* Cleanup

* move bounded-heap accumulator to utils

* switch to test.check testing

* Bad test just to see what our CI will report

* remove purposeful failing test (was for CI)

* collection.type -> collection.authority_level

* Test the actual ee scoring impl, not the wrapped one

locally i had enhancements enabled so the tests were passing, but in
CI it was failing as it did not have enhancements enabled