[sx4im]

What & why

Fixes #209

Two places in the docs pointed somewhere the binary did not go:

1. The README documents --vram-reserve-mib, and the engine accepts it, but strata.exe --help never listed it. Added the missing line to usage() in generate.cpp, next to --peer-reserve-mib. The tuning sweep reads its flag list from --help, so this also makes the flag visible there.
2. strata bench run exists only in the Terminal launcher (bin/strata), not in the engine binary, which answers unknown argument: bench. The protocol doc, the community guide, and the closing line of tools/strata_bench.py all said bare strata bench run with no hint which strata they meant. They now name the Terminal launcher and note the engine binary has no bench subcommand.

No README change: the capacity table was already right.

Area

• Engine, CLI or server (edge0-infer/)
• Model asset tools (edge0-infer/tools/assets/)
• Chat app examples (edge0-infer/examples/)
• Model docs (docs/)
• Other

How was it tested?

• Rebuilt the --help text from the usage() literals and ran the tuning flag-inventory regex over it: --vram-reserve-mib is now listed.
• python3 -m py_compile tools/strata_bench.py passes; pytest edge0-strata/tests/tuning/test_tuning.py passes (37 tests).
• The engine binary was not built here (needs the CUDA toolchain), so --help was verified from the source, not from a built binary. CI is the gate for that.

Checklist

• ctest --preset cpu -L "L0|L1" passes locally in edge0-infer/ (engine changes)
• Ran the higher test levels this change affects (see edge0-infer/docs/testing.md)
• Added/updated tests for new behavior
• Updated relevant docs / READMEs (incl. translations if user-facing)