Allow the duplicate_count column to be missing

[torognes]

When running sequence existence mode, it would be convenient to allow the duplicate_count column to be missing in the ’sequence’ file, and then just presume duplicate_count=1 for all sequences. Right now CompAIRR fails if the column is missing, but I think 99% of the sequence existence use cases would not consider duplicate counts on the sequence side, and may consider duplicate counts on the repertoire side (but to be safe/backwards compatible, we could still consider those counts for sequences if the duplicate_count column is provided).

[LonnekeScheffer]

Minor nuance: I think this doesn't need to only apply to sequence existence mode, but to any more; if the column is missing, just presume the duplicate_count = 1 for that file.

[LonnekeScheffer]

In practice I'm finding when we download data from the AIRR data commons, it may contain a combination of missing and present duplicate count values in one file (since a file can contain repertoires from multiple sources). I think it's safe to just use duplicate_count=1 as a default value for any missing duplicate count (even if they are specified for other sequences in the file).

[pavlovicmilena]

A bit related to this issue: if --ignore-counts is set, it would be nice not to check if the column exists or if it contains invalid values (eg, -1, '', ..). Now I get an error when trying to compute overlap if counts are -1, even though I am not using counts. In this case, a duplicate count of 1 can be assumed, as Lonneke says :)