Add option to redact/mask human reads

[fanninpm]

https://github.com/ncbi/sra-human-scrubber has the default behavior of replacing identified human reads with poly-"N" strings of equivalent length. Would you be interested in adding this to Deacon?

I imagine this would have follow-up questions regarding paired reads where one read identifies as human while the other read does not. I think that the current behavior discards the pair entirely if at least one read of the pair identifies as human, which makes sense for deletion of human reads.

[bede]

Apologies for delay. Would you mind explaining your motivating use case for this feature? I'll consider adding it if likely to be broadly useful. While the --summary report does not provide read IDs or individual lengths of depleted sequences, it does provide the depleted sequence counts and total depleted base pairs. You can invert the filter by removing the --deplete flag, but keeping both search and --deplete output requires two passes currently.