Problem: it is unclear when the tests are running NLP-related tests

[jrwdunham]

The functional tests are written in such a way that if foma and/or MITLM are not installed on the host, then certain tests will pass vacuously. On a TravisCI host, these OS deps are not installed. On a docker-compose local deploy, they should be installed. This is problematic if you want to know whether all of these tests are passing.

Proposed solution: add a strict flag to the test runner so that the test will fail if these foma/MITLM-dependent tests fail.