Moved changed files into a dropdown

Closes #29778

See merge request !13128

Don't bother going through an entire Banzai pipeline for empty text

See merge request !13300

This bails out of `cacheless_render` immediately unless the provided
text is present, since there's no point.

This is a slight improvement in our test performance. Across the
creation of 1,000 `Namespace` records, which caches the `description`
field and which is blank by default in its factory, this saves about
four seconds, which... sure. Why not.

Resolve "Specific Async Script Loading by using a Page Variable"

Closes #34904

See merge request !12759

Avoid plucking Todo ids in TodoService - take 2

See merge request !11415

Use mixin for new dropdown style

See merge request !13274

implement #inspect for all Referables

See merge request !13285

fix #3686 make tarball download url to end with extension

Closes #3686

See merge request !13178

Don't include EmailHelpers manually, pick with rspec

See merge request !13257

fix #35133 strip new lines from ssh keys

Closes #35133

See merge request !13234

Re-organise "issues" indexes for faster ordering

See merge request !13278

correctly scroll to element with sticky height offset

added specs to sticky util file
added `No files found.` text when results are empty

This makes navigating through diff files quickly. Currently we just
toggle a list, which could be pretty big. This moves it into a dropdown
to make it much easier.

Also includes a filter bar to quickly search for certain
files/extensions.

Closes #29778

Fixed sign-in restrictions buttons not toggling active state

Closes #35882

See merge request !13270

Migrate Gitlab::Git::Repository#count_commits to Gitaly

Closes gitaly#415

See merge request !13121

Merge request commits background migration

See merge request !12685

When trying to run an UPDATE, this query is ran:

```sql
UPDATE `todos`
INNER JOIN `projects` ON `projects`.`id` = `todos`.`project_id`
SET `todos`.`state` = 'done'
WHERE `todos`.`user_id` = 4
  AND (`todos`.`state` IN ('pending'))
  AND (EXISTS
         (SELECT 1
          FROM `project_authorizations`
          WHERE `project_authorizations`.`user_id` = 4
            AND (project_authorizations.project_id = projects.id))
       OR projects.visibility_level IN (10,
                                        20))
  AND `projects`.`id` IN
    (SELECT `todos`.`project_id`
     FROM `todos`
     WHERE `todos`.`user_id` = 4
       AND (`todos`.`state` IN ('pending')))
  AND (`todos`.`state` != 'done')
```

But MySQL does not like the subquery used to filter on
`projects.id IN (SELECT ...`

Because the subquery queries from the same table:

> Error: You can’t specify target table ‘todos’ for update in FROM clause

So as workaround, wrap it in another subquery, where the original
subquery is aliased using the `AS` statement.

Mostly inspired by https://stackoverflow.com/a/43610081/89376

TodoService should not call `.select(&:id)` on todos, because this is
bad performance. So instead use sub-queries, which will result in a
single SQL query to the database.

https://docs.gitlab.com/ee/development/sql.html#plucking-ids

Move API "basic usage" to be more visible

See merge request !13171

Update reply_by_email_postfix_setup.md, included starting courier-authdaemon after installation.

See merge request !13218

`:mailer` is needed to pick it easily, while
`type: :mailer` is needed for picking it automatically for
tests located in spec/mailers/*_spec.rb

It's a bit complicated in spec/services/notification_service_spec.rb
but we'll leave it alone for now.

fix Jump to next discussion

Closes #35232

See merge request !13076

By adding various composite indexes we can reduce the time spent
retrieving issue lists. Because of the way these indexes are built
column wise we can also remove some standalone indexes, keeping the
total number of indexes in check.

Make dropdown style on project page consistent

Closes #35778

See merge request !13193

Don't send rejection mails for all auto-generated mails

Closes #28472

See merge request !13254

Previously, we stored these as serialised fields - `st_{commits,diffs}` - on the
`merge_request_diffs` table. These now have their own tables -
`merge_request_diff_{commits,diffs}` - with a column for each attribute of the
serialised data.

Add a background migration to go through the existing MR diffs and migrate them
to the new format. Ignore any contents that cannot be displayed. Assuming that
we have 5 million rows to migrate, and each batch of 2,500 rows can be
completed in 5 minutes, this will take about 7 days to migrate everything.