Clean up IgBLAST parser

[Jason Vander Heiden]

The IgBLAST parser in MakeDb should read the column names of each hit table from the comments in the igblast output file, rather than basing column name off position.

Also, it's significantly slower than it needs to be, likely due to pandas which we can probably eliminate/optimize in someway.

[Jason Vander Heiden]

Parsing using column names instead of positions done in 626e6df.

[Jason Vander Heiden]

Lots of cleanup done in 2317cf. It's still kinda messy, but cleaned up enough to close this issue.