feat(memtrack): resolve allocator libs without scanning the whole nix store

[GuillaumeLagrange]

Dynamic discovery globs /nix/store/glibc/lib/libc.so.6 and gcc/libstdc++, matching every version in the store (each a distinct file, so dedup can't collapse them). memtrack then attaches uprobes to all of them even though the target loads one, which made a trivial run take minutes on a nix machine.

Add two opt-in escape hatches, checked before the glob:

• CODSPEED_MEMTRACK_ALLOCATOR_LIBS: explicit path-list of allocator libraries.
• CODSPEED_MEMTRACK_AUTO_LIBS: toggle that auto-resolves the libraries the target binaries (CODSPEED_MEMTRACK_BINARIES, already set by the runner) actually load, via each binary's own loader in list mode (honours nix RPATH).

Both fall back to the existing glob when nothing is set or resolved, so default behaviour is unchanged. On a nix box this cuts discovery from 25 attached allocator instances (>2 min) to 1 (~1.8 s).

[codspeed-hq]

Merging this PR will not alter performance

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

✅ 7 untouched benchmarks

---

_{Comparing memtrack-nix-speedup (facc889) with main (7ce2c98)}