fix(gitlab): add missing commit_updated_at column used by the MR incremental filter

[DoDiODev]

fix(gitlab): add missing commit_updated_at column used by the MR incremental filter

Summary

#8959 changed GetMergeRequestsIterator (backend/plugins/gitlab/tasks/shared.go) to
filter merge requests by

GREATEST(gmr.gitlab_updated_at, COALESCE(gmr.commit_updated_at, gmr.gitlab_updated_at)) > ?

but commit_updated_at does not exist on _tool_gitlab_merge_requests. There is no field
on the GitlabMergeRequest model, no migration script, and no code writing the value —
the only occurrences of the identifier in the whole repository are the three lines added
by that PR.

Consequently every incremental run of Collect MR Notes — the only caller passing a
stateful collector, mr_commit_collector.go passes nil — fails with:

subtask Collect MR Notes ended unexpectedly
  Error 1054 (42S22): Unknown column 'gmr.commit_updated_at' in 'where clause'

Approach

Rather than reverting #8959, this PR makes the intended behaviour actually work:

• add CommitUpdatedAt *time.Time to GitlabMergeRequest
• add migration script 20260812_add_mr_commit_updated_at.go (registered in register.go)
• maintain the column in ExtractApiMergeRequestsCommits: after extraction it is set to
  the latest commit_authored_date of the MR's commits, scoped to the current
  connection/project
• the filter in GetMergeRequestsIterator is left as introduced by fix(gitlab): always re-collect MR commits regardless of MR updated_at… #8959 and now resolves
  against a real column, so MRs that received new commits without their own updated_at
  being bumped (e.g. force-pushes) are picked up again

Testing

• make migration-script-lint
• go test ./plugins/gitlab/e2e/ -run TestGitlabMrCommitDataFlow — green on MySQL 8
  and PostgreSQL 14; the extractor's UPDATE ... SET ... = (SELECT MAX(...)) and the
  GREATEST/COALESCE filter behave identically on both engines
• go test ./plugins/ -run TestTableInfo

Notes

The GREATEST(...) expression returns NULL on MySQL when gitlab_updated_at is NULL,
so such MRs stay excluded — same as before #8959, where gitlab_updated_at > ? also
evaluated to NULL. No behaviour change there.