The absolute temperature input to the homogeneos freezing routine is not the most updated one

@PeterCaldwell @AaronDonahue

Folks, the absolute temperature `t_atm(k)` is used to identify the homogeneous freezing level, but the most recent updated temperature is the potential temperature (`th_tmp(k)`) rather than the absolute temperature:
https://github.com/E3SM-Project/scream/blob/1c3cb7dbb629898b218c7a8f36af910a1aecf2bf/components/eam/src/physics/cam/micro_p3.F90#L4118

You need either to update the absolute temperature at the same time as the absolute temperature (end of `_part2`), or just use the potential temperature to diagnose the homogeneous freezing level (`th_atm(k)*inv_exner(k)`).

Also, the latent heat variants (`_vapor`,`_sublim`,`_fusion`) are still constant based on the values at the beginning of the time step. Are these kept constant in other routines that SCREAM inherited from E3SM?
https://github.com/E3SM-Project/scream/blob/1c3cb7dbb629898b218c7a8f36af910a1aecf2bf/components/eam/src/physics/cam/micro_p3.F90#L1323