The comment for ORDER_HALF incorrectly stated it represents “Number of valid distinct x coordinates on the curve”. This is wrong - ORDER_HALF is simply half the group order, commonly used for ECDSA low-S signature normalization. The number of valid x coordinates on secp256k1 is approximately (p-1)/2 where p is the field size, which is a completely different value.
Also clarified the ORDER comment from “plus 1 for infinity” to “including infinity” for better accuracy.