Remove legacy layout and ship lock-free C#/C++/Python protocol

[rantri]

Summary

• remove the legacy StoreProfile/layout and make SMS2 the only supported protocol
• ship interoperable lock-free implementations for C#, C++, and Python
• replace legacy fixtures, documentation, packaging, samples, and validation with one canonical protocol
• harden lifecycle recovery, cross-process ownership, qualification diagnostics, and Linux per-store owner artifact isolation

Production qualification

Production source is frozen at fc605a0; 85e9124 only records the qualification result.

• Spec-Kit: 137/137 tasks complete; protocol checklist 30/30; requirements checklist 16/16
• Managed: 916 tests passed on Windows and 916 on Linux
• Native: 24/24 tests plus install and clean-consumer validation on Windows and Linux
• Python: 83/83 clean-installed-package tests on Windows and Linux, including rebuilt wheel/sdist and sample
• Interop: 153/153 normal plus 10/10 stress on each platform; all nine ordered runtime pairs, 1,000 values per pair, and 10,000 lifecycle cycles
• Docker: 24/24 native, 153/153 interop, and 10/10 stress
• Packaging/docs: NuGet clean consumer, documentation validation, retired-path inspection, and diff checks passed
• Linux owner-artifact regression: 12,000 unrelated root files and 64 concurrent cold opens stayed within the 500 ms per-open budget

The final verification uses the owner-approved compact delta policy: the prior immutable high-volume matrix is the baseline, while the exact production source reran the full managed/native/Python/interoperability/package suites and the focused lifecycle regression. The strict full-rollup contract remains documented for future releases.

Breaking change

There is intentionally no backward compatibility with the retired legacy layout. Existing consumers must close/drain old stores, recreate them under SMS2, and republish application-owned values.