Prevent resuming a saved preset

[ZenAlexa]

Closes #4204.

What changed

• refuse dstack preset resume once the session has a committed preset.yml; cleanup may leave session.json at running, interrupted, or failed
• let the saved preset own the list row, using the creation session only to decorate its progress, so a stale cleanup status cannot produce a second row for the same ID
• cover the service-level gate, the CLI behavior, and stale interrupted/failed display states

Why

PresetStore.save() atomically commits the verified artifact before run cleanup. Cleanup is already best-effort and cannot invalidate that artifact, so its interruption must not make the completed creation resumable. A genuinely interrupted session with no saved preset keeps the existing resume behavior.

Verification

After syncing current master into e54083b7, the preset tests use the current session_id field.

• Preset service and CLI tests: 353 passed, 2 skipped.
• Expanded CLI suite: 520 passed, 18 skipped, one failure in the unchanged metrics color-label test. The same test fails in an isolated rerun; its source and implementation match current master.
• Ruff and diff checks passed for the five contribution files.

The previous head passed Pyright with zero errors.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-08-30T06:16:04.768636Z	eeb4581	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[ZenAlexa]

The focused preset tests, Ruff checks, and Pyright pass on the current head. The fork workflow is action_required; please approve run 33296412298 so the full Test Build matrix can start.

[ZenAlexa]

Thanks for enabling the fork workflow. The full Test Build matrix is now green across Linux, macOS, Windows, Python 3.10–3.14, frontend, runner, gateway, schema generation, and lint; eeb45819 is ready for review.

[ZenAlexa]

I've synced this with current master and updated the regression fixtures for session_id.

The saved preset continues to own completion and its list row; the expanded preset tests cover that behavior on the updated branch.