As part of the plan to make title an optional attribute for Iteration,
the non-null constraint is removed along with the uniqueness constraints
from the database while the application-level checks are preserved.

We are including a down migration to set any null titles to their
primary keys (ids) because we want to start accepting null titles
behind a feature flag in the same release.

Changelog: changed

Generate unique name for security provider factory

See merge request gitlab-org/gitlab!82358

https://gitlab.com/gitlab-org/gitlab/-/issues/354821

Add prepend mod to wiki page

See merge request gitlab-org/gitlab!82165

Revert disabling multi project pipeline viz for free user

See merge request gitlab-org/gitlab!82329

Doc Consistency: user - authentication and authorization

See merge request gitlab-org/gitlab!81763

Add missing tier badge, change inline tier badges to links

Expose filter by `updated_before`/`updated_after` in GraphQL

See merge request gitlab-org/gitlab!81959

Fix broadcast message on subscriptions page

See merge request gitlab-org/gitlab!82183

Create new index for security ci builds with new features

See merge request gitlab-org/gitlab!82047

Add detail about multi-project specific runners

See merge request gitlab-org/gitlab!82148

Add unique index for security training providers

See merge request gitlab-org/gitlab!82235

Stop using UrlHelper includes in specs

See merge request gitlab-org/gitlab!82184

Migrate customers unavailable to shared haml alert

See merge request gitlab-org/gitlab!82325

Added more details about heading titles

See merge request gitlab-org/gitlab!82217

Based on a slack discussion, there are some titles we
should avoid. These guidelines should help us
focus our content and make it easier to find

Add missing demodulize call to class name

See merge request gitlab-org/gitlab!81879

The `BackfillJiraTrackerDeploymentType2` migration is remaining
stuck in pending
state at various customers because the job state update implementation
is passing the wrong string when sending a class name for lookup

The other migrations that do similar things
use the `demodulize` function
to transform a module name into just the unqualified classname but this
migration appears to have missed it

This change adds the missing method call to ensure future migration runs
(for yet-to-upgrade customers, or customers trying to re-run this) 
update the background migration jobs table correctly

Add frontend for connection section

See merge request gitlab-org/gitlab!82005

When the integrationFormSections feature flag is
enabled and `sections` is provided from the backend,
we now add a separate "Connection details" section.

For now, this section will include ActiveCheckbox and
some of the DynamicFields which have been filtered
by the backend.

We use dynamic imports here since not all integrations
have this sections defined yet and there will be more sections
that are integration specific (for example, Jira).

Log error message for report upload

See merge request gitlab-org/gitlab!82123

Secure Files UI MVC

See merge request gitlab-org/gitlab!79784

Migrate loading icon to be Pajamas compliant

See merge request gitlab-org/gitlab!81985

Migrated GPG key loading icon to the new `gl-loading-icon`

Changelog: changed

Annotate usages of ActiveRecord:Base in Pseudonymizer

See merge request gitlab-org/gitlab!81491

Allow Types::GlobalIDType[::WorkItem] to receive IssueID as input

See merge request gitlab-org/gitlab!81490

Update test user delete task and test user name

See merge request gitlab-org/gitlab!82162

- Gives the user a name identifiable as a QA test user.
- Includes the original name in the delete task.

Fix templates: "Category:X" is not a scoped label

See merge request gitlab-org/gitlab!78536

This reverts merge request !79889

Geo: Forward project repo pulls to the primary when repo is out of date

See merge request gitlab-org/gitlab!82034

A secondary site will now forward project repo pulls to the primary site
if the latest repo change is not yet replicated.

Current limitations:

We assume last_repository_updated_at is a timestamp of the latest
change. So we compare the timestamp of the latest sync to
last_repository_updated_at.

- Unfortunately, project wiki updates also touch
  last_repository_updated_at. So wiki updates will also bump the repo to
  out-of-date.
- Unfortunately, successive last_repository_updated_at touches are
  throttled within Event::REPOSITORY_UPDATED_AT_INTERVAL minutes. This
  is currently set to 5 minutes. So for example, secondary sites will
  think highly active repos are up-to-date after the first sync of each
  5 minute interval.

This is a rough improvement over the current code.

Next, we can iteratively improve the accuracy of
`.repository_out_of_date?` to fully address the underlying issue. It may
be rather involved, so this MR helps by reducing the scope of that
future change.