* Move *ignore-cached-results* to middleware

Had tried to adjust the binding to work, but the binding ended before
all of the bound functions and such were submitted to work, so when
that code actually executed the binding was no longer in
place. Annoying and it should be in middleware anyways

* Remove now unused import (cache became keyword in middleware not binding)

* skip some field logic when dealing with ExpressionDimensions

* hide xray drilldowns when there are expression dimensions

* fix tests by limiting filtering to just expression dimensions

* add a cy test

* update comment

* mild cleanup

* Add repro for #14652

* Fix blank screen when DB does not exist

DataSelector assumes `schemas` is an array.
But it's `undefined` if db does not exist

* Fix data model page state when DB does not exist

* Move prop-types to the top

* Fix invalid database request

* Do not break waterfall when there is a null

* links and collection checkbox image

* not on card movement

* managing people

* permissions

* auditing question table fields

#16693 Repro: Binning from a column popover (distribution) on explicitly joined field renders incorrect title (#16705)

#16675 Repro: UI inconsistencies for bucket display on tables from explicitly joined tables (#16703)

Localization also works already without reference to null semantic type, here is test to memorialize this (#16641)

It works, here's a test for it

Previous behavior searched only on columns backend name. This one searches on column frontend name (field name), table name, schema name - but not on all of them at once, just each of them separately

#16674 Repro: UI inconsistencies for bucket display on tables from implicitly joined tables (#16698)

* Add initial binning tests for saved QB question with explicit joins (via simple question)

* Add initial binning tests for saved QB question with explicit joins (via custom question)

* Add initial binning tests for saved QB question with explicit joins (via column popover)

* Update base question query

* Update tests for simple question mode

* Update tests for custom question mode

* Update tests for the column popover entry mode

* Fix comments

[Binning e2e coverage] Asserting on the bins from the saved QB question using implicit joins (#16662)

* Add initial binning tests for saved QB question using implicit joins (via simple question)

* Update tests for simple quesiton mode

* Add initial binning tests for saved QB question using implicit joins (via custom question)

Move block of code about getting input placeholder out of render function in FieldValuesWidget component (#16686)

* support date ranges for click behavior that updates dashboard filter parameters

* fix spec

* reorganize tests

* 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