CI - Move things to a small runner

[bfops]

Description of Changes

Move a bunch of jobs to the small runner:

•  Workflow / job                            VM hash    Duration    Peak CPU    Peak guest used    Min guest available    Peak QEMU RSS
  ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━  ━━━━━━━━━━  ━━━━━━━━━━  ━━━━━━━━━━  ━━━━━━━━━━━━━━━━━  ━━━━━━━━━━━━━━━━━━━━━  ━━━━━━━━━━━━━━━
   CI / Merge queue no-op/reuse             cc9dea2e          8s        1.8%           0.54 GiB               3.29 GiB         3.21 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   CI / CODEOWNERS check                    eef82464         24s        1.4%           0.64 GiB               3.18 GiB         4.08 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   CI / Check CLI docs                      775318c4      2m 00s       ~100%           2.35 GiB               1.47 GiB         4.09 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   CI / Check packages publishable          233a0c5f         19s       30.3%           0.79 GiB               3.03 GiB         4.08 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   CI / Check smoketest module list         a1d851c3          9s        0.7%           0.57 GiB               3.26 GiB         3.59 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   CI / Verify global.json symlinks         dbfc3ead         17s       14.0%           0.87 GiB               2.95 GiB         4.05 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   CI / Linux update-flow test              596ed296         38s       ~100%           1.34 GiB               2.49 GiB         4.08 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   CI / Build and test WASM bindings        623212de         33s       92.9%           1.11 GiB               2.72 GiB         4.07 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   CI / TypeScript tests                    3044428e      1m 29s       98.3%           2.24 GiB               1.58 GiB         4.08 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Check merge labels                       5f9965fb          3s        1.0%           0.53 GiB               3.30 GiB         2.54 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Git tree / Based on master               8f811de8          4s        0.8%           0.54 GiB               3.29 GiB         3.02 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Git tree / Check release dependencies    a1074f0a         27s        4.5%           0.73 GiB               3.10 GiB         4.08 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Internal Tests / coordinator             1b5166ad     40m 23s        4.7%           0.84 GiB               2.98 GiB         4.08 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Release dry run / Validate invoker       c7e0509d          4s       34.5%           0.62 GiB               3.20 GiB         3.04 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Release dry run / Build cargo-release    30c12ff9         19s       87.0%           0.99 GiB               2.83 GiB         3.99 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Release dry run / Release crates         5302db6e         10s        9.3%           0.66 GiB               3.16 GiB         3.34 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Release dry run / Release C#             7dc1430f         59s       71.0%           1.26 GiB               2.56 GiB         4.08 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Release dry run / Release C++            a911c40a         12s       29.1%           0.67 GiB               3.15 GiB         3.28 GiB
  ───────────────────────────────────────  ──────────  ──────────  ──────────  ─────────────────  ─────────────────────  ───────────────
   Release dry run / Release npm            fcad7030         56s       64.8%           1.81 GiB               2.02 GiB         4.09 GiB

We did have a few jobs max out their CPU, but they didn't run for very long so I think they're good on these small runners.

API and ABI breaking changes

None

Rollback safety impact

n/a

Expected complexity level and risk

1

Testing

• CI runs and passes
• The table above demonstrating that the jobs fit properly into this table

[jdetter]

Seems fine to me, make sure we merge + deploy the infra PR first 👍

[joshua-spacetime]

This seems a bit aggressive. Should we start with the jobs that are mostly I/O bound first before transitioning cpu bound ones? Notably cargo parallelizes compilation; pnpm does the same for installation. Any jobs that invoke those build paths could see a regression.

[bfops]

This seems a bit aggressive. Should we start with the jobs that are mostly I/O bound first before transitioning cpu bound ones? Notably cargo parallelizes compilation; pnpm does the same for installation. Any jobs that invoke those build paths could see a regression.

Which jobs are of particular concern to you? Notably, all the ones triggered by CI were measured here. Per my comment:

We did have a few jobs max out their CPU, but they didn't run for very long so I think they're good on these small runners.

I'm happy to move the release jobs and the high-CPU CI jobs back to large runners if you prefer.