ci: parallelise jobs at runner capacity >1, keep heavy jobs apart (closes #127) (#128)

## What & why

The runner's `capacity` was raised to 2. The six CI jobs have no `needs:` between them, so they already schedule concurrently now — this PR makes that safe and tidy rather than enabling it.

- **Keep the two memory-heavy jobs apart.** `verify-stack` now `needs: [mutation]` — not a data dependency, but so Stryker and the full-stack-bring-up + Playwright browser never run at once on the one host and re-trigger the e2e OOM (#126, commit d5e5fa2). `if: ${{ !cancelled() }}` keeps verify-stack running even when the mutation ratchet fails, so we don't lose its signal, while still honouring cancellation.
- **Light jobs stay dependency-free** (lint / build / unit / frontend) → they parallelise up to runner capacity.
- **Supersede stale runs** via a workflow `concurrency` group, so a new push cancels the previous run and frees the slot instead of piling up.

Net effect at capacity 2: the light jobs pair up (and overlap `mutation`), then `verify-stack` runs alone — shorter wall-clock, no heavy-heavy collision.

Closes #127

## Definition of Done

- [x] Linked issue (#127).
- [x] Conventional Commit referencing it.
- [ ] CI green — this PR **is** the test: it exercises `needs`, `if: !cancelled()`, and the `concurrency` group on Gitea. Watch that (a) verify-stack starts only after mutation, (b) verify-stack still runs, (c) the workflow parses (concurrency accepted).
- [x] No app/docs/ADR impact (CI-only).

## Notes for reviewers

- **One thing to watch on this first run:** if this Gitea version doesn't support the top-level `concurrency` key, drop that hunk — the `needs`/`if` guard is the load-bearing part and is plain job-graph syntax.
- **Cross-run collisions** (two different PRs' `verify-stack` on the 2-capacity runner) aren't controllable via intra-workflow `needs`. If that becomes a problem, the clean fix is a second runner (or a dedicated capacity-1 label for the stack job) rather than ordering — out of scope here.

Reviewed-on: #128

.gitea/workflows/ci.yaml

@@ -9,6 +9,12 @@ on:
 permissions:
   contents: read
 
+# Supersede stale runs: a new push to the same branch/PR cancels the previous run, so the runner's
+# concurrency slots aren't spent on commits nobody is waiting for (refs #127).
+concurrency:
+  group: ci-${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
 # Self-hosted runner — see docs/runbooks/ci.md for the runner setup.
 # `uses:` are absolute, tag-pinned URLs (CLAUDE.md §8.7 / §15).
 
@@ -129,12 +135,20 @@ jobs:
           path: services/bff/StrykerOutput/**/reports/mutation-report.html
           if-no-files-found: warn
 
-  # One stage for every check that needs the live stack. On the single self-hosted
-  # runner jobs run sequentially, so booting OpenZaak once (instead of once per job)
-  # is the cheapest layout (issue #58). No setup-dotnet: the ACL test runs in a built
-  # image and everything reaches services by container IP. Needs Docker + egress
+  # One stage for every check that needs the live stack. Booting OpenZaak once (instead
+  # of once per job) is the cheapest layout (issue #58). No setup-dotnet: the ACL test runs
+  # in a built image and everything reaches services by container IP. Needs Docker + egress
   # (base images, nuget, selectielijst.openzaak.nl).
+  #
+  # `needs: [mutation]` is NOT a data dependency — it serialises the two memory-heavy jobs so
+  # they never co-schedule now the runner has capacity >1. A concurrent Stryker run + full-stack
+  # bring-up + Playwright browser on one host is what OOMs the e2e (commit d5e5fa2, #126). The
+  # light .NET/frontend jobs have no `needs`, so they still parallelise up to runner capacity.
+  # `if: !cancelled()` keeps verify-stack running even when the mutation ratchet fails (so we don't
+  # lose its signal) while still honouring run cancellation from the concurrency group above.
   verify-stack:
+    needs: [mutation]
+    if: ${{ !cancelled() }}
     runs-on: ubuntu-latest
     steps:
       - uses: https://github.com/actions/checkout@v4