Stronger deterministic fixed node+fixed phase tests needed

[prckent]

Given #4763 we need a stronger test or set of tests on the fixed node and fixed phase condition, particularly at the deterministic and unit level. Attempts to introduce a similar bug in future should fail in CI.

One immediate possibility would be to increase the timestep in one of the batched DMC deterministic tests so that it has a high rejection rate.

It might be wise to formally add e.g. some of Umrigar's classic DMC calculations to the test set. Many of these have been run by hand over the years. These could also test timestep dependence on an ongoing basis.