Create a basic validator for a Babel run

[gaurav]

The goal is to have a program that we can point at a Babel run that runs some basic tests on the generated files, including:

• Testing whether all the expected compendia, synonym and report files exist and are larger than some minimum size
• Check if the same identifier shows up in multiple files
• Check that each file (except umls.txt) only contains a single Biolink type
• Check CURIEs (e.g. make sure we don't have CHEBI:CHEBI:41423 as per Remove CHEBI:CHEBI:41423 from PUBCHEM.COMPOUND:2662 results #69)
• Add a test for CHEBI.CHEBI just to make sure it's fixed TranslatorSRI/babel-validation#16
• Does every clique have a label?
• Is every UMLS ID in a clique somewhere?

We're also tracking this in the Babel repository at #36

Copied over from TranslatorSRI/babel-validation#14