ci: report sccache stats from the build server

[AnthonyRonning]

Research’s Unix cache statistics queried the default sccache server, while the builders selected a Unix socket under a temporary directory created by each Nix shell. That could report zero requests from a different server after a successful build.

Keep the build socket under Actions’ stable per-job RUNNER_TEMP and use a small reporting helper that shares the existing build configuration. Wire the nine existing desktop, Android and Rust Unix statistics steps through it. Explicit socket overrides and local temporary-directory fallback remain supported; Windows, compiler-cache keys, signing and runner selections are unchanged.

Validation:

• All 12 applicable aarch64-darwin flake checks passed.
• Normal pre-commit hook: workflow lint and all 433 repository tests passed.
• Actual isolated Rust compilations across separate pinned Nix shells: first query reported one miss; the next reported one hit and one miss, with zero cache errors. This verifies reporting, rather than application cache performance.
• Independent diff review found no blocking issues.

---

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

[github-actions]

Maple development preview: https://4ddd173e.maple-ca8.pages.dev

Commit: 63cfbd9ad9d9d01761aa913c06435d8571cb63e8

Uses development API, billing, flags and PCR configuration. Cloudflare Access applies.