FE-1211: Summarize completed execution comparisons

[kostandinang]

Why

After an execution comparison finishes, the operator can be left with a transcript filename or a verbose recap instead of a clear answer about whether the run completed and where its evidence lives.

What

Generate one final summary from the retained attempt records. The summary shows each lane result, explains invalid attempts, reports cleanup once, and provides absolute paths to the retained report and evidence. The comparison flow returns this summary automatically after the report is written.

How to test

1. Complete an execution comparison with both Brunch and Claude Code lanes.
2. Confirm the final response starts with Execution comparison complete and presents each lane in a separate result block.
3. Confirm an invalid lane includes a plain reason, while a valid lane does not show an invalidity reason.
4. Confirm cleanup appears once as Cleanup: done, without repeated cleanup clean labels.
5. Confirm every artifact path is absolute and opens the matching retained report, attempt, or oracle evidence.
6. Exit the Pi session and confirm the pi-session-*.html transcript is emitted after the comparison summary rather than replacing it.

[cursor]

PR Summary

Low Risk
Changes are limited to dev execution-comparison operator tooling and operator prompts; no production auth, data, or runtime paths.

Overview
Adds a summary path on execution-comparison-operator that reads retained attempt-records (after report.md exists) and prints a fixed "Execution comparison complete" block: per-lane validity, terminal outcome, command/browser checks, invalidity reason when needed, one Cleanup line, and absolute paths to report, attempts, and oracle reports.

The compare-execution Pi prompt now requires running that command after report.md and returning its stdout verbatim as the final operator response—no hand-written recap or transcript-only endings. Covered by new execution-summary tests and prompt contract tests.

^{Reviewed by Cursor Bugbot for commit 38c6bd9. Bugbot is set up for automated code reviews on this repo. Configure here.}

[kostandinang]

• FE-1211: Summarize completed execution comparisons #374 👈 (View in Graphite)
• next

This stack of pull requests is managed by Graphite. Learn more about stacking.

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 9a64e0c. Configure here.}