Consolidate lightweight workflow jobs for clarity and performance

[hdamker]

Problem description

The release automation reusable workflow currently has 17 jobs in the reusable workflow (19 including caller-level). Several are lightweight (2-6s compute each) but each requires its own runner startup (~3s gap). The high job count makes the workflow harder to understand, trace, and maintain:

• Data flow requires tracing across many needs: references and explicit output passing
• GitHub Actions UI shows a long list of jobs (most skipped per run)
• The job dependency diagram requires significant effort to keep readable

Proposed change

Combine 4 lightweight jobs into their natural parent jobs:

1. assemble-context → derive-state: Pure post-processing of derive-state outputs; shares the same tooling checkout and Python setup
2. validate-command → derive-state: Single github-script step (~20 lines); no separate setup needed
3. create-sync-pr → publish-release: Only runs on publish success; shares tooling checkout and Python; natural step-level condition

This reduces the reusable workflow from 17 to 13 jobs with no logic changes — purely mechanical refactoring.

Impact

Path	Jobs before	Jobs after	Time saved
/create-snapshot	5	3	~9s
PR merge	5	4	~3s
/publish-release	6	3	~12s

Primary benefit is architectural clarity (simpler needs: graphs, less output plumbing, cleaner Actions UI). Latency savings are modest but measurable (~3s per eliminated job boundary, based on 47 runs on ReleaseTest).

Additional context

• No functional changes — all consolidations are mechanical (move steps, update needs: references, rewire outputs)
• Detailed analysis with measured timings available in the technical documentation