Ignore remarks and empty end-of-verse paragraphs

[isaac091]

Since the draft remark only gets added to the draft, the vrefs were not aligned between the two files.

---

This change is 

[benjaminking]

Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @isaac091)

---

silnlp/common/compare_usfm_structure.py line 103 at r1 (raw file):

    tokenizer = WhitespaceMarkerTokenizer()

    gold_file_sents = [

It might be helpful to either add a comment or split this out into a separate method with a descriptive name, since what it's doing is a little unintuitive.

[isaac091]

I added another commit that is completely different from the original one but that is doing a similar type of thing as far as excluding elements that we don't want to factor into the score. A recent update to the USFM parser has it include the empty end-of-verse paragraph markers in row outputs, and that was artificially inflating the similarity scores.