AleksanderObuchowski's picture
Faster extract(): batched end pointer, shared schema tokens, length-sorted batches (~6x throughput, same outputs)
4679420 verified