Allow SQLServerAdapter To Opt Into A Few Tests

Update comment for Scrubber usage [ci skip]

Added test to ensure that we dont break #to_h again when trying to restore the speed-up from 26dd9b26 (cc: @fxn)

Revert "significant speedup of AC::Parameters#permit" [26dd9b26] as it broke Parameters#to_h on at least fields_for-style nested params.

This was preventing the test suite from being run in isolation

Add test for update_counters with empty touch array

`timestamp_attributes_for_updates_in_model` returns an empty
array when a model has no `updated_at` or `updated_on`.

So my previously thought uncommon case is a lot more likely
now.

This reverts commit a0a1ede8.

Make all rails commands work in engine

Fix UUID primary key with default nil in legacy migration

The exact inspect output of a BigDecimal is out of scope for what we're trying
to communicate about `dup` and `duplicable?` here.

Adding two examples distracts is disctracting, so keep the docs from before
since our minimal version is Ruby 2.2.2.

[ Koichi ITO, Jon Moss, Kasper Timm Hansen ]

This reverts commit 2163874d, reversing
changes made to 46fdbc52.

Closes #27683.

Seeing a code sample that leads to what we're guarding against:

```ruby
Topic.update_counters(1, replies_count: 1, touch: [])
```

It doesn't look like a case people would ever intentionally end
up with. Thus we're better off sparing the conditional.

Note: it could happen if a method returns an empty array
that's then passed to `update_counters` and its touchy friends.
But `[].presence` can fix that once people see their query blow
up.

[ Eugene Kenny & Kasper Timm Hansen ]

remove unused require

Several representation of BigDecimal has changed in Ruby 2.4.0+ [ci skip]

This reverts commit 66e5b9d3.

We're seeing some test failures in AR postgresql tests

I see no reason not to use the newest stable version.

In #26672 we blocked use of Ruby keywords as identifiers for view
locals but inadvertently broke the use of instance variable names
as identifiers. Whilst not explicitly documented this behavior has
been around for a long time and there's no need to break it now.

Fixes #27480.

remove unnecessary addition of `lib`

Fix broken sample code in action_view/helpers/output_safety_helper.rb…

`InlineAdapter` is not used from 1f8558fa.

cf. https://github.com/ruby/bigdecimal/pull/42

`TestTask` add `lib` directory by default.
Ref: https://github.com/ruby/rake/blob/master/lib/rake/testtask.rb#L39..L41

This is a regression test for a fix included in
https://github.com/rails/rails/commit/bad9bfbea6d6af9dc28583e08a49492668087393.

Without that change, this test would fail with:

    ActiveRecord::StatementInvalid: SQLite3::SQLException: near "WHERE": syntax error: UPDATE "topics" SET "replies_count" = COALESCE("replies_count", 0) - 1,  WHERE "topics"."id" = ?

instead, rewrite them to no-op

in case String or any other ancestor class' respond_to_missing? was defined.

Add the touch option to ActiveRecord#increment! and decrement!

in case Array or any other ancestor class' respond_to_missing? was defined.

`type_condition` should be overwritten by `create_with_value` in `scope_for_create`

`type_condition` should be overwritten by `create_with_value`. So `type`
in `create_with_value` should be a string because `where_values_hash`
keys are converted to string.

Fixes #27600.

Small spelling error

existing_acrnoyms -> existing_acronyms

If a process is forked more than once, the pool was grabbing the oldest
spec, not the most recent spec. This wasn't noticed before because most
folks are lilely forking the process only once.

If you're forking the process multiple times however the wrong spec name
will be returned and an incorrect connection will be used for the
process.

This fixes the issue by reversing the list of spec names so we can grab
the most recent spec rather than the oldest spec.

CI against JRuby 9.1.7.0