Unit tests not catching major problems

[danjampro]

While tests are passing with a relatively high code coverage, several major problems with basic usage were uncovered while testing on Huntsman last week. This includes problems with e.g.:

• The state machine
• Exposure timeouts
• Bisque mounts

A lot of this may be down to the POCS ethos of logging warnings rather than raising (and possibly catching) errors, combined with tests that gloss over the underlying issues and pass when they really shouldn't. I believe we either need to improve our tests or start raising more errors rather than warnings.