41 hits below threshold

[tyleraland]

Added a new argument that can optionally include hits that would otherwise be thrown out because they were below the best rank threshold.

I've included some notes describing what it looks like the classifier is doing and where I've made changes, in an effort to avoid a careless mistake:

Blast results get filtered several times

• if specimen map given, filter hits with unrepresented qseqid
• "raw_filtering", -- filter on coverage, min/max identity
• remove no_blast_results ... (hits with no sseqid) ... (added back in later)
• discard blast hits whose sseqids aren't in seq_info
• filter results not in taxonomy, add "tax_name", "rank" columns
• "join_thresholds" -- adds "threshold" columns for every rank, if applicable
• save current blast hits as results_belowthreshold; we will be sidestepping subsequent filtering steps and outputting filtered hits to a separate file *
• select_valid_blast_hits ... throws away sub-optimal rank hits ...
• drop unnecessary columns
• rename columns (tax_name_assignment, rank_assignment)
• "condense assignment hashes"
• "condense_ids"
• Add "rank_condensed" column, selecting the taxonomic rank for "condensed_id" taxid
• Add "star" column if one hit is 100%
• "assign" ... I think this is for classification name
• Add back no_blast_results ("qseqids")
• blast_results is done being filtered; determine which hits have been filtered and save them to a separate data frame to be output to a separate file