ci(replay): Overhead measurement

[vaind]

implements #6503

• sample comment: chore: replay metrics vaind/sentry-javascript#1 (comment)

Progress:

• metrics collection (Vitals, CPU, JS Heap)
• retry & multi-run logic
• current run analysis
• baseline run analysis
• implement replay test app
• run in CI
• add PR comments using GH APIs
• download baseline data from main branch using GH APIs

Outstanding/known issues

• A comment is not produced for PRs running from forks - this is due to action security policies. There is a non-trivial workaround so if it's necessary to have these metrics from fork-originating PRs, I can look into that too (as a follow up).
• I've seen (multiple times) that the app that includes replay somehow blocks UI and browser during shutdown - maybe this needs more investigation or is it a known limitation?

[Lms24]

Hi @vaind! Great to see progress on the overhead measurements!

I haven't looked too closely over this PR but I was just wondering if there's a specific reason you chose puppeteer over e.g. playwright? The reason I'm asking is that long-term, we'd like to remove Puppeteer in favour of Playwright as it's causing quite a few issues for us and we'Ve had good experiences with Playwright so far. Therefore I'd like to avoid using it in new projects we add to the repo.

I know that you're not yet finished but feel free to tag me or anyone from the team if you have questions. Cheers!

[vaind]

Thanks for having a first look @Lms24. As for Puppeteer - no specific reason, I've seen it used somewhere in the repo so went for it. Switched to playwright now 👍

[Lms24]

Thanks for addressing our feedback! Will merge it soon