the first query (q1) exactly matches the reference, and the NM tag has value zero.
the second query (q2) has one 2bp deletion and a single mismatch and the NM has value three, which would set Hit to have edits to three, so then the mismatches method would return two instead of the expected one.
I think we count a multi-base indel only once when subtracting the indels from the edits to obtain mismatches https://github.com/fulcrumgenomics/fgprimer/blob/6cf2542e927ced37dd0dce4c335de8dff07789c7/src/main/scala/com/fulcrumgenomics/primerdesign/offtarget/BwaAlnInteractive.scala#L94C7-L94C12
@tfenne thoughts?