[Profile #478] Add hardware-counter profiling for every canonical losing hot path

Parent: #478. Related: #479-#488 #498.

Wall-clock and Python/native attribution are not enough to distinguish compute, memory, branch and vectorization failures. Add Linux perf/PMU profiling where available and equivalent macOS sampling/counter evidence where practical.

Capture cycles, instructions/IPC, branch misses, cache misses, vectorization/code-shape evidence and hot symbols for each canonical losing operation, with a portable fallback when counters are unavailable. Normalize interpretation by work units rather than comparing raw counters across machines.

Acceptance: every current canonical loss has a saved hot-symbol/profile report; issues #479-#488 cite the dominant bottleneck class before optimization; post-fix profiles demonstrate whether the intended bottleneck actually moved.

Start with parent issue #478 and related issues #479-#488 and #498 to identify every current canonical losing operation and the evidence they already contain. Define the profiling and portable-fallback scope from the acceptance criteria, then verify that each loss has a saved hot-symbol/profile report, a cited bottleneck class, and post-fix evidence showing whether that bottleneck moved.