This fixes a number of CVEs in libgit2:
https://github.com/libgit2/rugged/compare/v0.28.3.1..v0.28.4.1

Add protected branch permission check to run downstream pipelines

See merge request gitlab-org/gitlab!20964

Enable no-jquery/slim ESLint profile

See merge request gitlab-org/gitlab!21775

Make use of uncached diffs where highlighting is not required

Closes #43654

See merge request gitlab-org/gitlab!21842

Add indexes on deployments to improve environments search

Closes #55353

See merge request gitlab-org/gitlab!21789

Previously it sounded like we'd always target the fork, regardless of
where the MR was opened.

Replace var with const / let in notes.js

Closes #36269

See merge request gitlab-org/gitlab!21845

(cherry picked from commit 0e4b14cd3b10f78a6a0686424a0ee0dc7b94c7a0)

Reuse prometheus chart values when editing alerts

See merge request gitlab-org/gitlab!21344

Fix init_create_cluster_spec timing out

See merge request gitlab-org/gitlab!21856

Limit max metrics embeds to 100

See merge request gitlab-org/gitlab!21356

Limits the maximum number of metrics embeds in a single GFM field
to 100 to help with loading. OVer 100 has a hard time loading.

Feature spec for cluster health board

See merge request gitlab-org/gitlab!20839

For some reason we need to create clusters_applications_prometheus
before stubbing Kube client discover request

To load diffs and other files, we make two SELECT calls in
`MergeRequest#environments_for`: one for the source branch, and one for
the target branch.

The source branch query makes a subquery:

```sql
SELECT environment_id FROM deployments WHERE project_id = 123 AND ref =
'source_branch'
```

The target branch makes this subquery:

```sql
SELECT environment_id FROM deployments WHERE project_id = 123 AND (ref =
'master' OR tag IS TRUE)
```

Both subqueries were taking over 250 ms to execute. When we add an index
for the deployments table on `(projects_id, ref)`, this execution drops
down to under 5 ms since the table usually has a small number of rows
for the source branch.

Adding a partial index on `projects_id WHERE tag IS TRUE` allows
PostgreSQL to perform two separate indexed scans: one for the ref, and
one for the tag condition. However, this query still takes 60 ms to run
for the gitlab-org/gitlab project because `master` has 10,000+ rows. We
should consider adding a LIMIT in a future iteration.

Closes https://gitlab.com/gitlab-org/gitlab/issues/55353

[ci skip]

[ci skip]

Generate sample Prometheus data

See merge request gitlab-org/gitlab!19987

Interpolate metric queries

Use yml for sample metric output to easily convert hash

Create a file for each metric

Replace metrics endpoint if using sample data

Need to create sample_metrics directory to run test

Moveing logic out of the controller

Insert live endpoint

Add specs for SampleMetricsService

Cleaning up controller specs

Return 404 instead of 500 when file is not found

**How?**
There's a dynamic loading in the test subject.
This causes the first spec to timeout past the
500ms threshold.

By eagerly loading this module in the spec file, the
module load time is no longer accounted for in
the spec time.

Implement bulk of Threat Monitoring store and filters component

See merge request gitlab-org/gitlab!21689

Part of [WAF statistics reporting][1].

This adds the bulk of the Vuex store for fetching environments and
the WAF statistics, and the filter bar component which allows selection
of the current environment.

This also:

- Aligns the UI with the latest designs, including the help link
  popover, copy, and the alert variant
- Fixes/improves existing tests (avoiding verbose snapshots)

[1]: https://gitlab.com/gitlab-org/gitlab/issues/14707

Alerts for the GitLab-managed app Prometheus are updated via helm.
Prior to this change, when alerts were updated, all helm values would
receive the current contents of `values.yaml`, with the `alerts` section
overridden, or falling back to the default chart values.

This meant that any time we made changes to `values.yaml` affecting
component versions (e.g. the versions of prometheus or alertmanager),
those would trigger a deployment of those component in the cluster,
potentially breaking the prometheus installation.

This was a very real problem, since we currently have hardcoded versions
of both prometheus server and alertmanager in `values.yaml`, and would
like to update the chart and remove the hardcoded versions in a single
change.

With this change, alert updates no longer take any values from
`values.yaml`; instead they only update alerts. We avoid reverting to
default chart values by using helm's `--reuse-values` flag instead of
the `--reset-values` flag.

See https://gitlab.com/gitlab-org/gitlab/issues/38294

Update Serverless docs for using an older version of gitlabktl

Closes #35655

See merge request gitlab-org/gitlab!21688

Fix padding structure on design comments

See merge request gitlab-org/gitlab!21839

Sync security branches using the Merge Train

Closes #38387

See merge request gitlab-org/gitlab!21647

Fix logs menu active state (cont)

See merge request gitlab-org/gitlab!21828

Allow pdf files to be opened on browser

See merge request gitlab-org/gitlab!21272

Allow styling links in broadcast messages

See merge request gitlab-org/gitlab!21522

Add reportSnippet permission to snippet GraphQL

See merge request gitlab-org/gitlab!21836

Add reportSnippet permission to the snippet
permmission types for GraphQL endpoints.

Add CI job to check karma jest migration

See merge request gitlab-org/gitlab!16336

Merge branch '35122-sentry-error-search-pagination-implement-pagination-in-sentry-errors-list' into 'master'

Implement pagination in Sentry errors list

See merge request gitlab-org/gitlab!21136

Resolve "Profile page remains active after user has been blocked"

See merge request gitlab-org/gitlab!21706

Add GraphQL mutation for changing weight of an issue

Closes #36317

See merge request gitlab-org/gitlab!21331

Enable forks removal upon visibility decrease by default

See merge request gitlab-org/gitlab!21834