* Revert "Clarify stacktraces from malli validation errors (#34712)"

This reverts commit 1a0f2d65.

* Alternative fix for mu/defn correctness issue

https://github.com/metabase/metabase/pull/34712 introduced a correctness
bug:

```clojure
(macroexpand '(metabase.util.malli/defn foo [x :- string?] :foo))

(def
 foo
 "Inputs: [x :- string?]\n  Return: :any"
 (clojure.core/fn
  ([a] (metabase.util.malli.fn/validate-input {:fn-name 'user/foo} string? a) ((clojure.core/fn [x] :foo) a))))
```

The simplest demo of the bug is

```clojure
user=> (metabase.util.malli/defn foo [x] a)
 #'user/foo
```

The macro used simple and predictable args for it's emitted
functions. This was not a problem with the original shape:

(different macroexpansions of `(mu/defn foo [x :- int?] (inc x))`):

prior to 34712:

```clojure
(def
 foo
 "Inputs: [x :- int?]\n  Return: :any"
 (clojure.core/let
  [&f (clojure.core/fn [x] (inc x))]
  (clojure.core/fn
   ([a] ;; simple argument name not a problem
    (metabase.util.malli.fn/validate-input
      {:fn-name (quote metabase.util.malli.fn-test/foo)}
      int?
      a)
    (&f a)))))
```

after 34712:

```clojure
(def
 foo
 "Inputs: [x :- int?]\n  Return: :any"
 (clojure.core/fn
  ([a] ;; simple argument name is now a problem
   (metabase.util.malli.fn/validate-input {:fn-name 'user/foo} int? a)
   ((clojure.core/fn [x] (inc x))
    a))))
```

after this PR:
```clojure
(def
 foo
 "Inputs: [x :- int?]\n  Return: :any"
 (clojure.core/let
  [&f (clojure.core/fn [x] (inc x))]
  (clojure.core/fn
   ([a] ;; again simple argument name is not a problem
    (try
     (metabase.util.malli.fn/validate-input
      {:fn-name (quote metabase.util.malli.fn-test/foo)}
      int?
      a)
     (&f a)
     (catch
      java.lang.Exception
      error
      (throw (metabase.util.malli.fn/fixup-stacktrace error))))))))
```

And the fix here is just to remove the StackTraceElements which
reference `validate`, `validate-input`, or `validate-output`.

* fix tests

tests are run in the user namespace similar to

```clojure
user=> (clojure.test/run-tests 'metabase.util.malli.fn-test)

Testing metabase.util.malli.fn-test

Ran 8 tests containing 43 assertions.
0 failures, 0 errors.
{:test 8, :pass 43, :fail 0, :error 0, :type :summary}
```

But we still want the namespace of _this_ file m.u.m.fn-test

This reverts commit 10cdb49d.

CI fails because of commit that is being reverted.

* audit-log only when hosted or has-feature: audit

* lint fix

* check for the right feature name, and test better

* fix metabase.api.setting-test

* fix a bunch of tests

* fix a couple more tests

* bit of cleanup

* lint fixes

* fix testing function

* fix whitespace

* linter fix

* fix so many tests

* fix the rest of the tests, IIUC

* fix the last test

* lint fix

* remove outdated tests (from the merge conflict)

* disable-other-sso-types needs to _reset_ premium features.

- patches in the audit-app feature by default.

* Just need to re-add :audit-app as a premium feature

* try improving disable-other-sso-types

Expressions which just wrap a real Field get `:id`, `:table_id` etc. so
they match a `[:field {} 7]` ref. If that real field exists (table
default or joined), prefer that over the expressions.

Fixes #33499

We added a bunch of OTEL spans to figure out what was going on with super long page loads. Turns out it was a slow pprint of massive objects.

This PR removes a few spans, but not all. The ones removed are either redundant or almost certainly not needed (I did remove the one around the offending code, but the negative space that would show if it happened again would reveal the issue -- LMK if we think we should put that one back).

One important thing to consider when reviewing this PR is if we like the current coverage level around the `metabase.events` code (plus a couple extra things that were instrumented). It feels right, but if it seems too onerous we could remove some.

As an observation, it _might_ make sense to add instrumentation to the high level functions in toucan2 itself, such as hydrate, select, etc. This would give us transparent instrumentation on the library and we could easily view traces to see the bounds of our db hot spots.

resolves #35880

* Fix "slugify collection url" flake

* Remove the `@flaky` tag

* try to restore the original position even when another card stole it

* avoid calling fetchCardData in undoRemoveCardFromDashboard for virtual dashcards

* test: e2e test for undo movement and card removal

* test to make the ci pass

* empty commit

* restore old scrollbehaviour

* refactored the magic number in a function so that also docs are in a single place

* Fix recent-views flake(s)

* Remove the `@flaky` tag

* Visit model query directly

* Use helper to save model metadata changes

* Limit the number of results even more

* Improve assertions

* Increase the viewport width in order to see all columns

Spans on stats for massive models are very long, on the order of 4s.

The bad span seems to encompass a small amount of logging and assertion
logic, including a with-out-str pprint, which I've found to be
notoriously slow for large objects. This PR slims down the logged object.

If this is the culprit we should see a substantial performance improvement.

* Adding more spans to understand stats `publish-event!`

https://stats.metabase.com/dashboard/2117

 still takes 4s to load and
the current spans don't seem to provide fine granularity around the
hot spot, but it seems to be happening in the early stages of
`publish-event!`. These extra spans provide finer granularity.

* Adding what is probably a non-needed span, but we want to over-instrument this, if anything. Once we id the hotspot we can remove what we want.

* add more spans

* remove :object from new spans

---------

Co-authored-by: Noah Moss <noahbmoss@gmail.com>

* [E2E] Fix dashboard card resize flake

* Remove the `@flaky` tag

* Add test to the slow group

* [MLv2] Put duplicate implicit join columns in separate column-groups

* Add tests for cards without id fields in metadata

* Fix linter

* Fix test

* Add Mantine `Modal`

* Add guidelines link

* Fix heading in storybook