Potential incomparable UMI's because of allowed extra characters

[veldsla]

When the UMI is parsed from the Illumina (compatible) read name the UMI is allowed to contain either + or _ characters. These are not removed from the sequence and when the layout of all UMI sequences in the BAM file is identical this is not an issue.

However when the UMI layout in the readnames varies across records this may leaf to comparing what are essentially incomparable UMI's.
We probably want to error-out of the program if this happens, although it seems a bit like an edge case.