IN filters of more than 16 values prune zones only by the set's overall min and max #101

SparkFilterToVortexExpression.convertIn (:157-204) turns an IN of up to 16 values (MAX_IN_DISJUNCTION) into an OR of equalities, which zone statistics can prune. Above 16 values it becomes a set lookup ANDed with the set's overall min/max, since "Vortex derives no zone bounds from a set".

So a 17-value IN keeps every zone between its smallest and largest value. On a file sorted by the filtered column, with values spread across the file, that's nearly every zone. A 16-value IN would have pruned all but the zones holding a value.

Example: a file sorted by a string key, read for about 80 keys spread through it. With the set lookup, almost every zone survives. With per-value pruning, only the zones holding one of the 80 keys would.

Proposal. Prune set membership per zone: keep a zone only if some value of the set falls within its [min, max]. That's a binary search over the sorted set for each zone, so large sets stay cheap. This likely belongs in the core expression's zone-statistics pruning rather than in the converter.