`TOOLS-CALLING-CORE` row records stale registry counts: `tool_parser_names()` 40 (actual 41) / `reasoning_parser_names()` 7 (actual 12)

[localai-bot]

What

The TOOLS-CALLING-CORE row in .agents/engine-matrix.md:156 records two
registry enumeration counts in its prose, and both are stale:

The name lists are ENUMERATED from the factories (tool_parser_names() 40
names / reasoning_parser_names() 7), not hand-written at the flag

Re-derived from the two enumerations on main today:

Enumeration	Row says	Actual	Anchor
tool_parser_names()	40	41	src/vllm/entrypoints/openai/tool_parsers/abstract.cpp:269
reasoning_parser_names()	7	12	src/vllm/entrypoints/openai/reasoning_parsers/abstract.cpp:72

Both counts were correct when they were written (2026-07-24,
CLAIM-DOCS-T2-FIXES). Names have been added under other rows since and the
prose was never moved with them:

• tool parsers 40 -> 41: muse_glimmer, 2026-08-10 under
  MODEL-MUSE-GLIMMER-W7.
• reasoning parsers 7 -> 12: olmo3, muse_glimmer, qwen3, mimo and
  minimax_m2_append_think landed under their own rows; docs/STATUS.md
  already reads 12.

Why this is worth an issue rather than a silent edit

The two halves belong to two different rows. The tool-parser count is
TOOLS-PARSER-BREADTH (#608) territory and the reasoning count is
SAMPLE-REASONING (#605); the row that CARRIES the stale prose is a third,
TOOLS-CALLING-CORE. Found while repairing the fresh-review findings on #643
(the TOOLS-PARSER-BREADTH W0 backfill), which is records-only on a different
row and deliberately does not touch this one.

Nothing gates this: the committed tests pin the counts in C++
(tests/vllm/entrypoints/openai/tool_parsers/test_detect.cpp:221 pins 41), so
the CODE is right and only the RECORD drifted. That is the same defect class
#608's W0 existed to repair, one row over.

Scope

Correct both counts in the TOOLS-CALLING-CORE row prose, re-derived from the
two abstract.cpp enumerations at the head that lands, in one change. No src/
or tests/ change: the enumerations are correct and already pinned by tests.

Found while implementing review findings on #643.