Should we change the distribution equality test in the test suite?

[axch]

Right now it uses either one-sample or two-sample chi^2, depending on whether one of the desired distributions is assessable. I recently found myself investigating the wisdom and appropriateness of this choice; results at https://github.com/probcomp/Venturecxx/blob/master/doc/on-comparing-discrete-distributions.md . Should we actually change anything?

[riastradh-probcomp]

For the record, the G-test is identical up to a multiplicative constant to what Jaynes calls the psi-test in Sec. 9.11 'Significance tests'.