Promote persistent MTP boundary split candidate default

[guelfoweb]

Promotes persistent MTP boundary split to default while keeping ORBIT_MTP_BOUNDARY_SPLIT=0 as rollback.

Review note:
This branch was built from a clean integration baseline that did not yet include the native backend/MTP package, so the PR may include native_llama, native_server, and MTP tests as part of the isolated merge candidate. The review-critical behavioral delta is limited to the persistent MTP boundary split default, rollback flag, and documentation.

Reviewer focus:

• src/orbit/native_llama/vendor/shim/orbit_persistent_mtp.cpp

  • boundary split is now default
  • ORBIT_MTP_BOUNDARY_SPLIT=0 restores previous behavior
  • fallback/guardrails remain in place

• docs/LLAMA_DEEP_CLIENT_EXPERIMENT.md

  • documents the new default, rollback flag, top1-greedy draft policy, and CPU-affinity benchmark requirement

• native MTP tests

  • validate persistent MTP and experimental MTP paths

Validation:

• PYTHONPATH=src python3 -m unittest tests.test_native_persistent_mtp tests.test_native_mtp_experimental -q PASS

• python3 -m compileall -q src tests scripts PASS

• git diff --check PASS

• smoke PASS on:

  • response_short
  • response_medium
  • long_prompt
  • session_reuse_repeat

• opt-out rollback verified with ORBIT_MTP_BOUNDARY_SPLIT=0

Benchmark:
Controlled with taskset -c 0-5, 10 runs per mode per scenario.

• response_medium: median -14.3%, p75 -14.5%
• long_prompt: median -16.3%, p75 -16.5%
• session_reuse_repeat: median -24.0%, p75 -24.9%
• response_short: median +0.5%, p75 +0.3%

Safety:

• zero assert
• zero unexpected fallback
• replay does not increase
• rollback flag works

Risk:
CPU-only latency remains sensitive to scheduler noise. Keep ORBIT_MTP_BOUNDARY_SPLIT=0 available during stabilization and run follow-up benchmarks on another host before removing the rollback flag.

Final decision:
Merge candidate.