[runtime] improve chain test failure diagnostics

[GsCommand]

Motivation

• The full-chain clean->summarize->classify->verify test could hit the 20s suite timeout without exposing which network call stalled, making CI failures silent and hard to debug.
• The change is test-only and intended to reveal the exact stalled step and handle a known validator warmup 202 path without touching runtime logic.

Description

• Added simple console.log statements before and after each network call in the chain (clean, summarize, classify, verify) so CI output shows the last completed step via runtime/tests/runtime-signing.test.mjs.
• Reduced request-level timeouts to 5s by using AbortController timeouts for both runVerb(...) and the final /verify?schema=1 helper to fail faster than the overall suite timeout.
• Introduced verifyReceiptWithTimeout(...) that mirrors runVerb error handling, returns { res, text, json }, and throws clear messages like verify failed: AbortError or verify failed: HTTP <status>: ... on errors.
• Implemented a one-time retry when /verify?schema=1 returns 202 with reason === "validator_not_warmed_yet", including logging the 202 response and waiting 1200ms before retrying, and ensured the test prints accumulated server stderr on failure.

Testing

• Ran node --test runtime/tests/runtime-signing.test.mjs and the test suite passed locally (all tests OK).
• The change is minimal and only affects test diagnostics and timeouts; no runtime or protocol contracts were modified.

---

Codex Task