fix(course-optimizer): fix start-analysis run-id assumption, polling, and not-started copy

[nsprenkle]

Description

Follow-up to #109. Three small fixes to the Course Optimizer extended report's start/poll cycle, found while testing #109 end-to-end against a devstack:

1. Stop expecting a run id from the start-analysis POST

In combination with edx/edx-platform#466 (which moved course export / compress to an async task), Studio's POST endpoint now returns 202 {"status": "pending"} immediately, before a run exists, rather than proxying the backend's real {run_id, ...} response synchronously.

This updates postCourseAnalysisReport to return void instead of { runId }, since nothing here actually consumed that value -- CourseOptimizerExtendedPage only ever used isPending/isError from the mutation.

2. Stop polling analysis status when no run exists

useCourseOptimizerReport's refetchInterval used to treat a null result (course has no run yet -- Studio proxies a 404 in that case) the same as an actively-pending run, and polled every 2s indefinitely -- hammering Studio with repeated 404s for a course whose analysis was never started. It now only polls while a run actually exists and is in an active status (PENDING/RUNNING/PARTIAL).

This is a real behavior change from what #109 originally shipped (and from what this PR's own first commit above assumed): starting a run in the same browser tab still works immediately, since useStartCourseAnalysisReport's onSuccess explicitly invalidates the query rather than relying on the interval. The one tradeoff: a run started in a different tab/session for the same course won't be picked up here until the page is reloaded, since polling no longer runs speculatively while there's no run to check on.

3. Clarify the never-scanned state vs. a load error

The "no run yet" state was a single muted line of body text with no heading (notStartedHeading was defined in messages.ts but never actually rendered), easy to mistake for an error state at a glance. Now renders a clear heading ("No previous scans found for this course") and points directly at the scan button, kept visually distinct from the real load-error state (isError).

Supporting information

Companion PR: edx/edx-platform#466

Testing instructions

1. npm test -- src/optimizer-page should pass.
2. npm run types should pass.
3. Manually, with both companion PRs applied: for a course with no prior run, confirm the status endpoint is called once (not every 2s). Click "Start analysis" -- it should transition to Running (polling resumes at ~2s) and eventually show the full report, then stop polling once Complete.

[Copilot]

🟡 Changes recommended

Polling can stop after an initial 404 before the asynchronously created run becomes available.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

Updates the Course Optimizer frontend for Studio’s asynchronous 202 Accepted analysis-start response.

Changes:

• Removes the unused run ID from the POST API.
• Updates polling, invalidation, and related tests.
• Revises pre-analysis messaging and headings.

File summaries

File	Summary
src/optimizer-page/extended/messages.ts	Updates no-run messaging.
src/optimizer-page/extended/data/apiHooks.ts	Adjusts asynchronous startup polling and invalidation.
src/optimizer-page/extended/data/apiHooks.test.tsx	Updates polling and mutation tests.
src/optimizer-page/extended/data/api.ts	Makes the start API return void.
src/optimizer-page/extended/data/api.test.ts	Updates POST API tests.
src/optimizer-page/extended/CourseOptimizerReportBody.tsx	Adds the no-run heading.
src/optimizer-page/extended/CourseOptimizerReportBody.test.tsx	Verifies updated no-run content.
src/optimizer-page/CourseOptimizerExtendedPage.test.tsx	Updates queued-start test fixtures.

Review details

• Files reviewed: 8/8 changed files
• Comments generated: 1
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.