fix(ci): start the Helm release after post-release housekeeping

[thomasqueirozb]

Summary

Start the Helm chart release only after post-release housekeeping has pushed to master.

The Helm release ends with the Kubernetes manifests refresh, which pushes to master. Today it starts in parallel with housekeeping. Housekeeping records master when it starts and refuses to push if master moved while it was generating, so a manifests push that lands during that window fails housekeeping. The automatic unfreeze then sees the failed Release Suite and keeps the freeze active, which needs a housekeeping re-run and a manual unfreeze dispatch.

With prepare-helm-release depending on housekeeping, the manifests refresh always starts from a master that already contains the housekeeping commit. If housekeeping fails, "Re-run failed jobs" on Release Suite also re-runs the Helm dispatch, so the chain, including the automatic unfreeze, completes without manual steps. The Helm release starts about 10 to 15 minutes later.

References

Related: #26538

Vector configuration

NA

How did you test this PR?

• Reproduced the race in a full sandbox release (vectordotdev/ci-sandbox, ci-sandbox-helm-charts, ci-sandbox-homebrew-brew): housekeeping recorded master at 20:59:26, the manifests refresh pushed at 21:03:40, and housekeeping failed with "master moved since housekeeping was generated". The sandbox Helm CI skips kind and ct install, so the Helm chain finished much faster than in production.
• actionlint .github/workflows/release.yml passed.

Does this PR include user facing changes?

• Yes. Please add a changelog fragment based on our guidelines.
• No. A maintainer will apply the no-changelog label to this PR.