Benchmark the real FastCoorFromBitBoardRank8ToRank1 call-site shape

Measures the actual ctzll+BIT_NUMBER_TO_COOR+bit-clear chain generate.c's
per-piece scan loops use, plus the bare BIT_NUMBER_TO_COOR macro alone,
to isolate the conversion arithmetic from the bit-scan around it.
Confirms (alongside an objdump check of the real binary) that LTO
already inlines these across translation units, so eval.c's many
remaining call sites against the "slow" extern CoorFromBitBoardRank8ToRank1
wrapper cost nothing extra in practice.

Co-Authored-By: Claude Sonnet 5 <[email protected]>

0 files changed, 0 insertions, 0 deletions