compile dataset should generate aggregate counts

[rlskoeser]

To simplify analysis, it would be helpful if the compile dataset added counts to the ppa and poem metadata files so we can easily view higher-level information.

ppa work metadata should include:

• number of excerpts
• number of poems
• number of poets

poem metadata should include:

• number of excerpts
• number of ppa works with excerpts
• number of pages with excerpts
• length of poem (characters; lines?)

[rlskoeser]

I'm assuming raw counts for now even though we know there are problems with deduplication on both poem and poet

[rlskoeser]

merged in #260, will deliver results via google drive