fix(action): retry transport errors when downloading the release archive

[blairham]

What

Add --retry-all-errors (and raise --retry 3 → 5) to the two curl calls in the install step.

-curl -fsSL --retry 3 -O "${base}/${archive}"
-curl -fsSL --retry 3 -O "${base}/checksums.txt"
+curl -fsSL --retry 5 --retry-all-errors -O "${base}/${archive}"
+curl -fsSL --retry 5 --retry-all-errors -O "${base}/checksums.txt"

Why

A consumer's CI run failed on its default branch with:

curl: (35) Recv failure: Connection reset by peer
##[error]Process completed with exit code 35.

curl --retry N only retries transient HTTP responses (5xx, 408, 429) and timeouts. A connection reset mid-transfer is a transport-level error, so curl exits 35 immediately and never touches its retry budget — --retry 3 was never going to cover this failure mode.

The run was otherwise green (test + build both passed); nothing was wrong with the repo or its hooks, and a plain re-run went green with no change. This step is pure infrastructure — fetch a tarball, verify its checksum — so an intermittent network fault should cost a retry, not a red check on someone's main.

--retry-all-errors requires curl 7.71+; GitHub-hosted runners ship 8.x.

Notes

• Worst case adds ~31s per download before a real failure (curl's backoff doubles from 1s).
• The checksum verification below is untouched — a truncated or corrupt download still fails loudly rather than being retried into a false pass.