Reimplement and rename get_annotation function

[standage]

This update replaces the get_annotation function with the new collapse_tandem_repeat and collapse_all_repeats functions.

I tried several ways to clean up the regex-based splitting approach, but this approach leads to unintuitive off-by-one errors for internal tandem repeats (i.e. 10 empty strings representing 11 copies of the repeat). Correcting for this resulted in opaque code any way I sliced it.

So I started from scratch and went with a recursive function that finds the first instance of the repeat in the sequence, collapses it, and then calls itself to repeat the process on the remaining sequence. This approach is a bit more concise and clear, and (most importantly) it works perfectly as a drop-in replacement for get_annotation!

I also added some doctests to the docstrings of the two new functions, so I updated make test in the Makefile to find and run these doctests.

[standage]

Ready for review and merge!

[standage]

Also, renamed the loci_need_split_anno function to collapse_repeats_by_length. Made some minor changes to variable names, but otherwise left that function alone.

[rnmitchell]

Ok, are you done changing things? I didn't want to merge it if you were still working on it.

All the tests passed and I used the ~1300 sequences we have "known" bracketed annotation for and all matched as well.

[standage]

Ok, are you done changing things? I didn't want to merge it if you were still working on it.

Well, I'm on a roll. 😀 Since you're actively responding, let me get one more change in there. Then it will be mergeable.

[rnmitchell]

hahaha exactly why I asked!

[standage]

Ok, this last commit renamed the split_string method to sequence_to_bracketed_form, updated all the places in the code where this was called, and changed a few long variable names to get everything to fit on one 99-character line.

[rnmitchell]

NOW are you ready? ;)

[standage]