fix(opencode): support pull_request_review as a trigger event

[argszero]

Issue for this PR

Closes #50247

Type of change

• Bug fix
• New feature
• Refactor / code improvement
• Documentation

What does this PR do?

pull_request_review was described as a supported event, but it was not actually handled. It was missing from USER_EVENTS, so it fell through to the generic "Review this pull request" prompt, and the instruction was read from payload.comment.body — a field a review payload does not have. A review whose body says /oc therefore did nothing.

This adds pull_request_review (types: [submitted]) to the event routing, to the generated workflow's on:/if: blocks, and reads the instruction from review.body for review events.

Review bodies can be null when a review is submitted without text, so the new resolveTriggerBody helper keeps two cases apart: undefined means the event has no body source at all (malformed payload worth failing on), while "" means an empty body, which is reported as a missing mention. The helper is a pure function next to extractResponseText, exported through github.ts like the other shared helpers.

github/README.md and the docs event table are updated to match the workflow opencode github install now generates.

How did you verify your code works?

• bun test test/cli/github-action.test.ts → 23 pass / 0 fail, including 6 new resolveTriggerBody cases (comment body for comment events, review body for reviews, a review never reading comment.body, a comment never reading review.body, body: null → "", no body source → undefined)
• The rest of the CLI test suite passes locally and tsgo --noEmit is clean
• Cross-checked the payload shapes against @octokit/webhooks-types: every PullRequestReviewEvent variant types review.body as string | null, and SimplePullRequest carries number, so the PR-context path resolves correctly

Screenshots / recordings

N/A — no UI change.

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR

[github-actions]

The following comment was made by an LLM, it may be inaccurate: