* wait for schemas/tables to load before running permissions grid logic

* don't return grid when given schema's tables are unavailable

I thought the collections appearing in the list page problem was solved, but this turned out to be a bug with the ttest. Upon actual inspection the lists remained.

To fix this without changing all sorts of crap that depended upon collections api contract (we futzed with the output but the futzing of the output is transparent to the frontend because of the entity thing they got going on), needed a way to elicit null response from collections endpoint. shoved one in. then renamed it because the first name, "dummy", makes it all sound kind of jank, so 'no-model' for explicitly asking for no model without handing in an empty set, which turns into the set of all models.

This is an 80% unjankening. Full unjankening would entail whacking this special case.

This mimics the situation in the notebook (custom question). A minor
difference, hence the omission of the previous implementation, is that
notebook breakout selection only allows one table at a time (the other
tables are "folded" in the accordion list component). In the case of a
simple question, every breakout selection happens in a sidebar, a long
list of all tables and all dimensions. Thus, it needs to support
multiple selection at once, which this commit finally implements.

* Test can't share dashboard without cards

* Fix can't share dashboard with markdown cards only

* Fix public dashboard crash while loading

* Remove duplicated test

* Add `aria-disabled` prop to Triggerable

* Subscribe to dashboards only if it has data cards

* Fix can't unarchive question without data access

* Test API checks perms when unarchiving a card

* Add initial binning tests for saved SQL (via simple question)

* Add more binning tests for saved SQL (via custom question)

* Add more binning tests for saved SQL (via column popover)

* Simplify the query

* Update visualization for custom question, time series

* Update comments with the references to the created issues

* Fix funnel appearance

* fix wide funnels get trimmed

* Fix incorrect question variable links

* Extract method building question URL

* Unskip the repro

* Add `isPersonalCollectionChild` utility

* Hide permission control for personal collections

* Hide permissions modal for personal collections

* Handle missing `collection.effective_ancestors`

* Use `sidebar` cypress selector

* fix trend visualization is not visible if the first column is not date

* skip repro

* review fix

* don't apply parameter filters to object queries

* unskip #8055 repro as well

* add units to new columns, add hover state for column headers

* fixup! add units to new columns, add hover state for column headers

* hide loader when filtering or sorting audit table

* fix tests

No tests caught this because of compounding errors on top of a generic
solution that already worked:

for h2, defined unix-timestamp->honeysql on [:h2 :millisecond] but
:millisecond is not a valid dispatch key, it is always
milliseconds (plural). And it was duplicated forgetting to use
:microseconds.

But this all worked because there are defaults in the
sql/query_processor

```clojure
(defmethod unix-timestamp->honeysql [:sql :milliseconds]
  [driver _ expr]
  (unix-timestamp->honeysql driver :seconds (hx// expr 1000)))

(defmethod unix-timestamp->honeysql [:sql :microseconds]
  [driver _ expr]
  (unix-timestamp->honeysql driver :seconds (hx// expr 1000000)))

```

so h2 was always using these fallbacks with the division:

```sql
"SELECT \"PUBLIC\".\"INCIDENTS\".\"ID\" AS \"ID\",
        \"PUBLIC\".\"INCIDENTS\".\"SEVERITY\" AS \"SEVERITY\",
        timestampadd('second', (\"PUBLIC\".\"INCIDENTS\".\"TIMESTAMP\"/1000000),--note division
                     timestamp '1970-01-01T00:00:00Z') AS \"TIMESTAMP\"
 FROM \"PUBLIC\".\"INCIDENTS\" LIMIT 1048575"
 ```

rather than the h2 optimized form:

```sql
"SELECT \"PUBLIC\".\"INCIDENTS\".\"ID\" AS \"ID\",
        \"PUBLIC\".\"INCIDENTS\".\"SEVERITY\" AS \"SEVERITY\",
        timestampadd('microsecond', --note microseconds
                     \"PUBLIC\".\"INCIDENTS\".\"TIMESTAMP\",
                     timestamp '1970-01-01T00:00:00Z') AS
                     \"TIMESTAMP\"
FROM \"PUBLIC\".\"INCIDENTS\" LIMIT 1048575"
```

The end result is that the defmethods were never called and the
default was used. So a bug, but not one that was visible. Good catch
jeff

Merge `39.4` into `master`

Merge `release-x.39.x` including ae6e9d9f without squash merging

* Move test inside a describe block

* Add repros for #16379

* keep click_behavior settings when reset cards settings

* unskip repro

* review fix

* add seriesIndex back to clicked object

* use transformed series when handled viz click

* unskip repro for related issue

* add additional test for this bug

Fix typo in DS name for app DB

Add test assertion

* Add new 'binning' e2e group to the CI

* Add initial set of tests for the binning options (buckets)