Suggestion

According to https://github.com/typescript-eslint/typescript-eslint/actions/metrics/usage, our biggest total minutes in the last month are:

1. ci.yml (linux): 17,478 minutes across 425 runs
2. ci.yml (windows): 7,610 minutes across 207 runs
3. semantic-breaking-change-pr.yml: 613 minutes across 359 runs
4. semantic-pr-titles.yml: 282 minutes across 282 runs
5. a11y-alt-bot.yml: 201 minutes across 201 runs

17478 / 425 = >40 minutes of CI time per commit. That's quite a lot! We've started to see more GHA throttling recently. Jobs/runs have been getting stuck waiting to start -> then for GHA runners to free up.

Filing this issue as a general one: what can we do to reduce the throttling in CI?

Additional Info

These more specific issues are split out of or generally related to this one:

• migrate from markdownlint to remark-lint #3223: using ESLint for Markdown linting would reduce the number of jobs per commit by one
• Repo: Onboard OctoGuide #11203 tracks combining semantic-pr-titles.yml and a11y-alt-bot.yml. That should improve a teeny bit for reducing our number of runs per commit / event.
• Repo: Per-job CI Install step takes a minute #11205
• Repo: Speed up semantic-breaking-change-pr.yml #11206

💖