Test the Gibbs kernels with the Geweke test

[emilyfertig]

This consists of multiple tests that follow the same basic structure:

• Take a single sample from an HIRM
• Incorporate that sample into a new HIRM
• Call the relevant transition method multiple times
• Use the Geweke test (Add a utility that implements the Geweke test #123 ) to test stationarity/sampling efficiency

Test each of these components

• Transitioning entity cluster assignments, using SRS as a metric (Add a test utility that implements SRS #124 )
• Transitioning relation cluster assignments, using SRS as a metric
• Transitioning latent values

The tests should be applied both to a case where the individual Gibbs kernel is transitioned repeatedly (with everything else held constant) and to a case where the whole model is transitioned.

[alex-lew]

I haven't done this kind of testing before but I think it's intriguing.

Is the idea to somehow use the SRS within the Geweke test, or as a separate way of testing the kernels for cluster assignments?