By creating and detaching partitions using the same lease, we can
easily skip the partition sync code if another node is currently
executing it. This is safe because we have six months of headroom for
future partitions and the job to sync partitions runs every six hours

Adds the partition_pruning_dry_run feature flag.

Even though the partition manager does not actually detach any
partitions yet, putting the logic for determining the set of partitions
to detach behind a flag makes rolling out this code safer, because the
logic changed touches the logic for creating new partitions.

Feature flag issue: https://gitlab.com/gitlab-org/gitlab/-/issues/335315

The previous code was using try_obtain once and two seperate locks,
so concurrent code would skip the critical section.

Switching to using in_lock and a single lock ensures that all concurrent
runners get to execute the code, and that partition creation and
partition detaching do not interfere with each other.

This worker runs `PartitionManager.detach_partitions` on a schedule to
trigger the detach logic for old database partitions that are no longer
needed

Enumerate the partitions that are outside of the retention period, and
for each one log a message that we are planning to detach it.

This way we can run this code in production for a while
and be sure that the choices of partitions to detach are correct.

This is the first step in providing support for detaching and dropping
old partitions of time-series database tables.

This identifies the set of partitions that should be detached by
calculating the set of desired partitions, and removing partitions that
are entirely before the retain_for cutoff date.

Resolve "Show warning if no .git in the URL when importing Repo by URL"

See merge request gitlab-org/gitlab!65739

Enhance image extension to support image uploads

See merge request gitlab-org/gitlab!65763

Clean up review app by namespace

See merge request gitlab-org/gitlab!63292

Prevent Transfering a Group with a Subscription

See merge request gitlab-org/gitlab!65689

Use `let_it_be` in filtered search feature specs

See merge request gitlab-org/gitlab!65976

Fix user popovers in design management discussions

See merge request gitlab-org/gitlab!66027

Update runner metrics definitions

See merge request gitlab-org/gitlab!64103

Only call fetch list items in graphql boards when list is not collapsed

See merge request gitlab-org/gitlab!65381

Account for skip_true_up flag in License validation

See merge request gitlab-org/gitlab!65793

Create download blob viewer

See merge request gitlab-org/gitlab!65915

Reset memoized bloat estimation upon #reset

See merge request gitlab-org/gitlab!65833

Translate GitLab

See merge request gitlab-org/gitlab!65148

Add copies of ArchiveTraceWorker and BuildFinishedWorker

See merge request gitlab-org/gitlab!64934

Migrate the deprecated "urlParamsToObject" to "queryToObject"

See merge request gitlab-org/gitlab!65894

Changelog: added

Add file copy ability to code search results

See merge request gitlab-org/gitlab!65749

Check PermitDataCategories.product_intelligence_enabled in submit service ping service [RUN ALL RSPEC] [RUN AS-IF-FOSS]

See merge request gitlab-org/gitlab!65998

Bump chemlab to allign with selenium-webdriver and rake version

See merge request gitlab-org/gitlab!65557

GithubImporter: Count fetched/import objects globally and by project

See merge request gitlab-org/gitlab!65700

[RUN ALL RSPEC] [RUN AS-IF-FOSS] Add collected_data_categories to usage ping payload

See merge request gitlab-org/gitlab!65336

Add project iteration cadences

See merge request gitlab-org/gitlab!64051

Enable GraphQL Runner API by default (runner_graphql_query)

See merge request gitlab-org/gitlab!65942

Add sitespeed-result to lint ignore

See merge request gitlab-org/gitlab!66013