I've found the skewness of delegation methods between `except` and
`only` in a88b6f25.

The `only` method is closely similar with `except` as `SpawnMethods`.

https://github.com/rails/rails/blob/e056b9bfb07c4eb3bcc6672d885aadd72bec574f/activerecord/lib/active_record/relation/spawn_methods.rb#L53-L67

It is preferable both behaves the same way.

This makes to ease testing `QUERYING_METHODS`.

Although the old name had a certain persistence, this ain't the kind of
file we're in now.

Foreign keys could be created to the same table.
So `remove_foreign_key :from_table, :to_table` is sometimes ambiguous.
This allows `remove_foreign_key` to remove the select one on the same
table with giving both `to_table` and `options`.

This change adds a new method that loads the YAML for the database
config without parsing the ERB. This may seem odd but bear with me:

When we added the ability to have rake tasks for multiple databases we
started looping through the configurations to collect the namespaces so
we could do `rake db:create:my_second_db`. See #32274.

This caused a problem where if you had `Rails.config.max_threads` set in
your database.yml it will blow up because the environment that defines
`max_threads` isn't loaded during `rake -T`. See #35468.

We tried to fix this by adding the ability to just load the YAML and
ignore ERB all together but that caused a bug in GitHub's YAML loading
where if you used multi-line ERB the YAML was invalid. That led us to
reverting some changes in #33748.

After trying to resolve this a bunch of ways `@tenderlove` came up with
replacing the ERB values so that we don't need to load the environment
but we also can load the YAML.

This change adds a DummyCompiler for ERB that will replace all the
values so we can load the database yaml and create the rake tasks.
Nothing else uses this method so it's "safe".

DO NOT use this method in your application.

Fixes #35468

Since #23461, all adapters supports prepared statements, so that clears
the prepared statements cache is no longer database specific.

Actually, I struggled to identify the cause of random CI failure in
#23461, that was missing `@statements.clear` in `clear_cache!`.

This extracts `clear_cache!` to ensure the common concerns in the
abstract adapter.

Adds a method to ActiveRecord allowing records to be inserted in bulk without instantiating ActiveRecord models. This method supports options for handling uniqueness violations by skipping duplicate records or overwriting them in an UPSERT operation.

ActiveRecord already supports bulk-update and bulk-destroy actions that execute SQL UPDATE and DELETE commands directly. It also supports bulk-read actions through `pluck`. It makes sense for it also to support bulk-creation.

Most existing tests expects `connection_handlers` has only one default
handler, but the test added at #34779 dirties that.

We need to reset `connection_handlers` to default in that case.

Closes #35471.

* Add `ActiveRecord::Base.connection.truncate` for SQLite3 adapter.

SQLite doesn't support `TRUNCATE TABLE`, but SQLite3 adapter can support
`ActiveRecord::Base.connection.truncate` by using `DELETE FROM`.

`DELETE` without `WHERE` uses "The Truncate Optimization",
see https://www.sqlite.org/lang_delete.html.

* Add `rails db:seed:replant` that truncates database tables and loads the seeds

Closes #34765

All the tests have a substantial chunk of identical setup effort (in
cases/helper.rb); this makes isolated tests run much faster, without any
change to the variety of how we load files.

In MySQL, the default collation is case insensitive. Since the
uniqueness validator enforces case sensitive comparison by default, it
frequently causes mismatched collation issues (performance, weird
behavior, etc) to MySQL users.

https://grosser.it/2009/12/11/validates_uniqness_of-mysql-slow/
https://github.com/rails/rails/issues/1399
https://github.com/rails/rails/pull/13465
https://github.com/gitlabhq/gitlabhq/commit/c1dddf8c7d947691729f6d64a8ea768b5c915855
https://github.com/huginn/huginn/pull/1330#discussion_r55152573

I'd like to deprecate the implicit default enforcing since I frequently
experienced the problems in code reviews.

Note that this change has no effect to sqlite3, postgresql, and
oracle-enhanced adapters which are implemented as case sensitive by
default, only affect to mysql2 adapter (I can take a work if sqlserver
adapter will support Rails 6.0).

Fixes https://github.com/rails/rails/pull/33611#discussion_r261549790
Related to https://3.basecamp.com/3076981/buckets/24956/chats/12416418@1631552581

[ci skip]

Follow up #35455, there are two more test cases unnecessary `if current_adapter?(:OracleAdapter)`

```ruby
$ ARCONN=oracle bin/test test/cases/associations/eager_test.rb -n test_include_has_many_using_primary_key
Using oracle
Run options: -n test_include_has_many_using_primary_key --seed 62842

.

Finished in 50.280024s, 0.0199 runs/s, 0.0398 assertions/s.
1 runs, 2 assertions, 0 failures, 0 errors, 0 skips
$
```

```
$ ARCONN=oracle bin/test test/cases/migration/index_test.rb -n test_add_index
Using oracle
Run options: -n test_add_index --seed 52034

.

Finished in 13.152620s, 0.0760 runs/s, 0.0000 assertions/s.
1 runs, 0 assertions, 0 failures, 0 errors, 0 skips
$
```

Since https://github.com/rails/arel/pull/337 Oracle adapter uses better
top N query using `fetch first n rows only`, which can remove this
unless condition.

* This commit passes with Oracle database
```ruby
$ ARCONN=oracle bin/test test/cases/relation/delete_all_test.rb -n test_delete_all_with_order_and_limit_deletes_subset_only
Using oracle
Run options: -n test_delete_all_with_order_and_limit_deletes_subset_only --seed 1081

.

Finished in 8.068626s, 0.1239 runs/s, 0.6197 assertions/s.
1 runs, 5 assertions, 0 failures, 0 errors, 0 skips
$
```

* SQL statement includes `ORDER BY` and `FETCH FIRST n ROWS ONLY`

```sql
  Post Destroy (12.5ms)  DELETE FROM "POSTS" WHERE "POSTS"."ID" IN (SELECT "POSTS"."ID" FROM "POSTS" WHERE "POSTS"."AUTHOR_ID" = :a1 ORDER BY "POSTS"."ID" ASC FETCH FIRST :a2 ROWS ONLY)  [["author_id", 1], ["LIMIT", 1]]
```

Kind of reverting https://github.com/rails/rails/commit/3a1cbc5c3b3bcb2de4be6e4469bb87b99759dc59

```
From: test/cases/date_test.rb @ line 26 :

    21:
    22:     invalid_dates = [[2007, 11, 31], [1993, 2, 29], [2007, 2, 29]]
    23:
    24:     valid_dates.each do |date_src|
    25:       topic = Topic.new("last_read(1i)" => date_src[0].to_s, "last_read(2i)" => date_src[1].to_s, "last_read(3i)" => date_src[2].to_s)
 => 26:       binding.irb
    27:       assert_equal(topic.last_read, Date.new(*date_src))
    28:     end
    29:
    30:     invalid_dates.each do |date_src|
    31:       assert_nothing_raised do

irb(#<DateTest:0x0000556618194668>):001:0> topic.last_read.class
=> Date
```

Refer rsim/oracle-enhanced#845
rails/rails#25897

```ruby
$ ARCONN=oracle bin/test test/cases/migration/columns_test.rb -n test_rename_column_with_multi_column_index
... snip ...

F

Failure:
ActiveRecord::Migration::ColumnsTest#test_rename_column_with_multi_column_index [/home/yahonda/git/rails/activerecord/test/cases/migration/columns_test.rb:113]:
--- expected
+++ actual
@@ -1 +1 @@
-["i_test_models_hat_style_size"]
+["index_test_models_on_hat_style_and_size"]

```

Kind of reverting #9395
Refer https://github.com/rsim/oracle-enhanced/pull/1703

This was using a model without a default scope what made the test don't
break anymore if the code is removed.

#35360 allows table name qualified if `from` has original table name.
But that is still too strict. We have a valid use case that `from` with
INDEX hint (e.g. `from("comments USE INDEX (PRIMARY)")`).
So I've relaxed the table name detection in `from` to allow any
extension like INDEX hint.

Fixes #35359.

`require "bcrypt"` is unsed since #15431.
`require "models/topic"` is unused since 893c647d.

Related and follows d333d852.

This is an alternative of 65c4b1b5.

https://buildkite.com/rails/rails/builds/59147#117e9445-23e8-455a-b486-ea0ae9636405/120-129

Some untransactional tests (e.g. `CallbacksOnMultipleActionsTest`) makes
`topics` table dirty.
We should reset dirty `topics` table before `SerializedAttributeTest` is
run.

https://travis-ci.org/rails/rails/jobs/499719624#L1209-L1215

It's mentioned everywhere as `ActiveRecord::RecordNotFound` so to be
coherent with the rest of the documentation I've applied it here.
Also doc was saying if the parameter is integer it coerces it which is
other way around.

This fixes both #34555 and #34738.

Revert "Merge pull request #34900 from gmcgibbon/fix_test_find_only_some_columns"

This reverts commit ff807f82, reversing
changes made to 9f1a07af.

Revert "Merge pull request #34560 from gmcgibbon/fix_decorate_leak_on_serial_attr_test"

This reverts commit bd623893, reversing
changes made to ec66c6a2.

Revert "Fix unstable `test_serialized_attribute_works_under_concurrent_initial_access` test"

This reverts commit 65c4b1b5.