perf(rallocator): consolidate v1 and Seismograph improvements

[Ralf Biedert (ralfbiedert)]

Summary

Consolidates the Rallocator v1 performance work and Seismograph snapshot/monitor improvements: medium-span caching and reclamation, eligible in-place reallocation, remote-accounting improvements, independently owned snapshot storage, and an offline snapshot viewer.

Public API changes

• Exposes rallocator::tuning_telemetry behind the existing tuning-telemetry feature, with explicit session control and allocation-free counter observations.
• Adds PeakLiveBytesScope, Stats::peak_live_bytes_scope, and optional lifetime/sample peak accessors. Rallocator reports aggregate-query sample peaks, not lifetime high-water marks.
• Requires allocator statistics section version 2. Legacy version 1 statistics are rejected; backward compatibility is intentionally not retained.
• Adds the documentation-hidden integration hook seismograph::snapshot::snapshot_collection_active(); no snapshot_arena_active() alias is retained.
• Adds seismograph view <SNAPSHOT-FILE> for read-only interactive inspection of saved captures.
• Propagates Cachet's Seismograph feature to BytesBuf when that optional dependency is present.

Existing allocator entry points and configuration remain unchanged. Eligible allocations can resize in place through the existing GlobalAlloc::realloc interface.

Validation so far

• Windows: 8,211 all-feature tests passed (4 skipped), excluding only the unchanged msvc_spectre_libs package because this machine lacks Visual Studio Spectre-mitigated CRT libraries.
• Windows: all-workspace no-default-feature tests passed (6,904 passed, 4 skipped).
• Workspace doctests passed: 1,262 with all features (same Spectre-package exclusion), 1,057 with default features.
• Rust 1.95 tests for all 11 changed crates passed: 1,314 all-feature and 1,261 default-feature tests.
• All 36 each-feature builds passed; Rallocator all-target/all-feature debug and release builds passed.
• Workspace Clippy passed with the same Spectre-package exclusion. Formatting, spelling, README, license, dependency-policy and branch-crate unused-dependency checks passed.

Draft while validation continues. The complete local just anvil-pr gate stops at the missing Spectre libraries; coverage, runtime-analysis, mutation, Linux and ARM results are not yet all confirmed. Hosted CI is the next source of actionable failures.

[github-advanced-security]

CodeQL found more than 20 potential problems in the proposed changes. Check the Files changed tab for more details.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.0%. Comparing base (ec1f6bc) to head (325a3ad).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##             main     #764     +/-   ##
=========================================
  Coverage   100.0%   100.0%             
=========================================
  Files         640      648      +8     
  Lines       85528    91129   +5601     
=========================================
+ Hits        85528    91129   +5601     

Flag	Coverage Δ
linux	99.9% <100.0%> (-0.1%)	⬇️
linux-arm	99.9% <100.0%> (-0.1%)	⬇️
scheduled	?
windows	100.0% <100.0%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.