Since migrations for EE models don't have factories in CE we need to
skip this cop in foss-only CI.

This commit also simplifies the implementation and improves the
performances be caching files operations.

If an association defines a scope with arity != 0
then the scope is not evaluated when preloading
that relation. This fixes it.

Changelog: fixed

This is to make it easier to support origin remote in the following
commit. Also we're trying to move logic out of Runner as it's become a
bit unmanageable.

In real usage if you call `ENV.fetch('NON_EXISTENT_KEY')` it will raise
a `KeyError`. But our previously implementation of of `stub_env` with
`nil` is causing it to return `nil` rather than raising an error. This
can cause us to write tests that don't match the actual production
behaviour. We might forget to rescue the exception in a certain place or
we also may not be able to write a test case for the expected behaviour.

```
irb(main):001> ENV.fetch('NON_EXISTENT_KEY')
(irb):1:in `fetch': key not found: "NON_EXISTENT_KEY" (KeyError)

irb(main):002> ENV['NON_EXISTENT_KEY'] = nil
=> nil

irb(main):003> ENV.fetch('NON_EXISTENT_KEY')
(irb):3:in `fetch': key not found: "NON_EXISTENT_KEY" (KeyError)
```

- add bad format and enforce pattern.

The new class will be responsible for all shell/commands execution

Now that the housekeeper creates an empty branch and merge request
before committing (so that the merge request URL can be used in keeps),
we need to skip filtered changes earlier to prevent pushing a branch
unnecessarily.

Adds a placeholder for the merge request URL which keeps can use in
their code, which will be substituted for the actual URL when the
merge request is created.

To enable this, the housekeeper now pushes an empty branch and
merge request before committing to the branch.

Signed-off-by: Rémy Coutable <remy@rymai.me>

Signed-off-by: Rémy Coutable <remy@rymai.me>

Signed-off-by: Rémy Coutable <remy@rymai.me>

Signed-off-by: Rémy Coutable <remy@rymai.me>

These special characters break the branch name. We might want to use
file paths and namespaced class names in `identifiers` in future and as
such we should strip these out when generating the branch name.

As suggested in
https://gitlab.com/gitlab-org/gitlab/-/merge_requests/144559#note_1772656778
it would be helpful for people to have a link to somewhere they can
provide feedback for housekeeper MRs. In future we might consider
automating the creation of a feedback for every keep. For now I don't
want to add extra friction for creating a keep so I've just created a
single meta issue. We'll see how we go with that before we decide if we
need more issues.

As suggested in
https://gitlab.com/gitlab-org/gitlab/-/merge_requests/144559#note_1772656778
it would be helpful for reviewers of
[`gitlab-housekeeper`](https://gitlab.com/gitlab-org/gitlab/-/tree/master/gems/gitlab-housekeeper)
MRs to know which specific keep generated the MR. This will make it
easier for them to locate the code and make improvements if they need
to.

This API was changed and we forgot the readme.

Changelog: other

These special characters break the branch name. We might want to use
file paths and namespaced class names in `identifiers` in future and as
such we should strip these out when generating the branch name.

Delete the specs for BufferedIO class

Update any other references in the codebase for V2

This adds logic to print the newly created (or updated) merge request
URL. This will be a nice shortcut for the user to jump to the MR and
double check it.

Since we're already practicing using "once-off" keeps and considering I
can imagine people will want to do this in the future I've added a small
section to the Readme about it.

This fixes the keep `OverdueFinalizeBackgroundMigration` which was
previously getting an error because `reviewers` is nil. Instead of
handling `nil` all over the place we can just set our arrays to default
empty values if they are optional. This also simplifies callers as they
can just append to arrays if it's cleaner.

During local usage of this we found it is often
[desirable to get the housekeeper to generate a specific MR](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/139747#note_1741757815)
and ignore the other ones. Since it just does them in order we had to
previously manually hack the keeps to skip the changes we don't care
about.

This change introduces a command line option `--filter-identifiers`
which takes a comma separated list which will be matched against the
list of "identifiers" in a change yielded from a keep. The identifiers
is an array so to allow for flexible regex matching of specific elements
in the array we filter such that all regexes must match at least one of
the identifiers in the list.

This `-r keeps/...` is proving to be unnecessary and likely it's fine to
just always load all the files in `keeps/*.rb`.

I considered the possibility of loading everything in `keeps/*.rb` as
well as keeping the option around in case someone wanted to manually
load a keep that was in a different directory but I couldn't think of
any good reason to do that. So to simplify the options I've just removed
it.