fix(cms): fold column index in queryFrequencyFast to match insert (non-pow2 cols)

[zzylol]

Problem

In sketches/CountMinSketch/CountMinSketch.go, the insert path folds the per-row column index back into range when cols is not a power of two:

c := int((hash >> shift) & s.mask)
if c >= s.Cols { c %= s.Cols }   // InsertWithHash, ~L229-231

but the query fast path queryFrequencyFast (~L354-357) indexed RowSlice(r)[c] with no fold. Because hashLayoutForCols rounds the mask up to the next power of two, a masked index can exceed cols, so querying a non-power-of-two CMS could panic (index out of range) and/or read the wrong cell. queryFrequencyFast was the only unfolded read site (estimateMatrixHash, QuerySum, QuerySum2 already fold).

This also matters for cross-language parity: the Rust asap_sketchlib CMS folds on both insert and query, so Go was the odd one out for non-pow2 cols.

Fix

Apply the identical fold in queryFrequencyFast so insert and query always land on the same cell.

Test

TestCMS_NonPowerOfTwoCols_QueryFold (cols = 17 and 2000, known counts):

• Before: panic: index out of range [26] with length 17 at CountMinSketch.go:357.
• After: PASS — QueryWithHash == FastEstimateWithHash, estimates ≥ true counts.

go build ./... clean; go test ./... passes (pre-existing CocoSketch / cross_language-needs-XTEST_DIR failures are unrelated, confirmed on untouched main).

🤖 Generated with Claude Code