docs(linkage): cave#316 — record the sharpened species findings in the baked-arm commentVerified

No behavior change (baked arm stays on the B.1 name-keyed fallback for
non-raw cells).  Comment now records what the session established: the
offending value is a genuine 2-cap ADR-0028 closure (the arity RAISE,
not not-a-function, proves lowtag 011 + widetag 0x08 — refuting the
cons-%CLOSURE hypothesis), gensym entry expecting 2 args not 4,
bisected to recycle-symbol's dolist/mvb capture region, reached
through a mis-installed baked-name cell.  Full repro recipe + the
make-program-error breakpoint plan are in cave#316.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>