fix(ci): repair benchmark failures in protocol and relay suites

[cursor]

Summary

Fixes the cargo bench CI job that was failing on PR #145 (and would fail on any PR touching benchmarks).

Root cause

1. protocol bench: chunk/write_read_roundtrip panicked on Ok(0) from chunk_read, but fragmented messages (4096-byte payload, 128-byte chunks) legitimately return Ok(0) until all continuation chunks are consumed.
2. relay bench CI: The workflow passed --test-threads 1, which is a libtest flag that criterion does not accept.

Changes

• Continue the read loop on intermediate Ok(0) returns; only panic when the buffer is exhausted without a complete message.
• Remove --test-threads 1 from .github/workflows/benchmarks.yml.

Testing

• cargo bench --bench protocol --features default -- --sample-size 10 --noplot
• cargo bench --bench relay --features default -- --sample-size 10 --noplot

  

---

^{Need help on this PR? Tag /codesmith with what you need. Autofix is disabled.}

---

Note

Low Risk
Benchmark and CI-only changes with no production runtime behavior.

Overview
Restores a green cargo bench CI job by fixing the protocol round-trip benchmark loop and the relay benchmark workflow flags.

The chunk/write_read_roundtrip loop no longer treats every Ok(0) from chunk_read as failure; it only panics when Ok(0) happens with an empty buffer, so fragmented large payloads (128-byte chunks) can finish across multiple reads.

The benchmarks workflow drops --test-threads 1 from the relay cargo bench invocation because that flag is for libtest, not Criterion.

^{Reviewed by Cursor Bugbot for commit 0e9504b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_2e2abc6d-58b8-4354-a9e1-911296b6d0ad)