Fix: restart & continue kills the live process, then refuses to replace it

[UsryAce]

Summary

• `restartWithModel(a, a.model, { resume: true })` — called by both "restart & continue" buttons (the per-agent row and the god/Michael row) — never sets `resumeOptional`.
• When the recorded session can't actually be resumed (stale/missing sessionId, or no matching rollout transcript for a codex-provider agent), giveUpOnResume either throws immediately (no sessionId case) or the flow has already killed the live pty before discovering the resume target doesn't exist (requireResume case) — either way the agent ends up dead with no replacement process started.
• The fallback this needs already exists and is already wired up one control over: the per-agent model-switch dropdown passes resumeOptional: true specifically so giveUpOnResume falls back to a fresh, non-resumed spawn instead of throwing. This PR wires the same flag into the two restart buttons.

Repro (what I hit)

1. An agent's process becomes unresponsive (in my case, a codex-provider agent whose session transcript never got flushed before it died).
2. Click "restart & continue".
3. Error: Existing session could not be resumed; no replacement process was started. — and the agent is now dead with nothing running, requiring a full app restart to recover it (which has its own issues — see the companion PR for #single-instance-hard-exit).

Fix

Two one-line changes, both purely additive (passing an already-supported option):

- onClick={() => restartWithModel(a, a.model, { resume: true })}
+ onClick={() => restartWithModel(a, a.model, { resume: true, resumeOptional: true })}

Test plan

• npm run typecheck passes
• npm run build succeeds
• Maintainer/CI: manual restart-a-hung-agent smoke test, since I wasn't able to get a locally-repackaged build to launch cleanly in my own environment to verify end-to-end (packaging issue on my end, not a concern with this specific diff — it's a 2-line, no-behavior-change-when-resume-succeeds change to an option that's already used elsewhere in the same file).

🤖 Generated with Claude Code

Before

Clicking "restart & continue" on a hung Codex-provider agent produced this error, with the process left dead and nothing replacing it (captured live tonight while diagnosing this exact bug):

Existing session could not be resumed; no replacement process was started.

This happened consistently, every time, on multiple different agents (confirmed on both a codex-provider agent whose session transcript never flushed, and separately after manually clearing a stale sessionId from registry.json, which changed the error to "No recorded session ID; current process was left running." - same outcome, still dead, still no replacement).

After

Same button, same class of agent, after applying this fix (resumeOptional: true on both call sites) to a live install and rebuilding: clicking "restart & continue" on a Codex-provider agent (Darryl) completed cleanly - agent card shows idle with no error text, ctx no status tick yet (i.e. a genuinely fresh session), zero residual error state. Repeated on a second agent with the same result.

[github-actions]

🚫 This PR is missing its before/after evidence

Every pull request here has to show its work. Screenshots or a short screen recording, before the change and after it.

• Before — no image or video under that heading
• After — no image or video under that heading

How to fix it: edit the description, keep the ### Before and ### After headings from the template, and drag an image or video under each. GitHub uploads it inline. This check re-runs the moment you save.

A bug fix with no visible surface still needs it: show the failing behaviour, then the same steps passing. A terminal recording is fine.

Genuinely nothing to show — a CI tweak, a typo, a dependency bump? A maintainer can apply the no-visual-change label. Please don't ask unless it truly has no observable effect.

📖 CONTRIBUTING.md → Evidence is mandatory