Compare original tokenization to subsequent modifications

[bwaldon]

• Addresses tokenized string showing up after server restart does not match saved tree #91: once an annotation is accepted, the latest tokenization is displayed to the user when the user subsequently views the sentence.
• Adds 2 fields to the database: senttok and origtok. senttok is a string which reflects the sentence tokenization after any manual edits; origtok is a string reflecting the tokenization prior to any manual edits.
• Introduces a "Revert tokenization" button in annotate mode. If the user clicks this button, the sentence is re-tokenized according to the original tokenization specified in the input .csv file (the file that SENTENCES points to in settings.cfg).

[nschneid]

Just to make sure I understand the rationale, the idea of storing the original tokenization in the database is that a user might accidentally modify the tokenization, save a tree, and wish to revert?

[bwaldon]

That's right. The original tokenization is also available in the SENTENCES .csv, so for sentences in that file, it's not strictly necessary to store the original tokenization in the database. But looking ahead to #4 (direct sentence entry), we'll need to store information about annotated sentences that come from somewhere other than that .csv. This change anticipates that future functionality (in progress at https://github.com/bwaldon/activedop/tree/direct_entry).

[bwaldon]

Following convo 11/19: going to revert changes that reference "original tokenization"

[bwaldon]

I'm actually going to close this PR because I think this is functionally identical to #92!