tooling: per-alloc-site memory tracker in compare_ffi_memory.rs

[polaz]

Context

Current compare_ffi_memory.rs reports total bytes allocated by Rust vs
FFI on each scenario but doesn't attribute allocations to specific call
sites. When the Rust path uses 1.4× more memory than donor we can't tell
which struct/buffer is responsible without instrumenting by hand.

Goal

Per-alloc-site attribution: track allocations through a custom global
allocator (or dhat-style sampling) so the report shows:

Site (file:line)	Bytes (Rust)	Bytes (FFI proxy)	Δ
frame_compressor.rs:142 (Vec::with_capacity)	8 MB	4 MB	+4 MB
...	...	...	...

Acceptance criteria

• Allocator wrapper or dhat integration that captures call stacks
• Report sorted by abs delta vs donor proxy (donor bytes inferred from
  ZSTD_estimateCCtxSize / direct ZSTD_sizeof_CCtx if accessible)
• Gated under a cargo feature (default off — adds overhead)
• Documented in compare_ffi_memory.rs how to enable / interpret output

Priority

Low — tooling for future memory regressions. Pull when the next memory
gap surfaces; don't pull preemptively.

Files involved

• zstd-cli/src/compare_ffi_memory.rs (existing harness)
• zstd-cli/Cargo.toml (new optional dep: dhat or custom allocator)

References

• Existing harness: zstd-cli/src/compare_ffi_memory.rs

[coderabbitai]

🔗 Related PRs

#43 - test(bench): expand benchmark parity matrix [merged]
#107 - perf(bench): add wildcopy candidate research bench and dashboard range filter [merged]
#143 - bench: add real peak-alloc metric for full encode/decode/memory baseline [merged]
#149 - feat(bench): #99 Rust↔FFI sequence-stream comparator [merged]
#168 - perf(huff0): #167 cheap entropy proxy for table_log selection — no FSE-encode per candidate [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!