Sometimes, the Akasaka solver has issues. This is usually caught and
then it is retried with a new omega. However, sometimes it goes bad
because the J matrix is singular, and this is not caught because the
error is not recalculated.

This commit recalculates the error term to prevent a bad result.

During saturated PHSU flash calculations, SatL and SatV states have an
imposed phase. This is good for stability, but there is a small chance
that they can both up up with a matching third variable (e.g.,
pressure) that is not actually at the saturation point. This commit
forces a final DT update without this requirement. If an actual solution
has been found, the the error term will still be small. If not, then we
throw an exception and try again.

This continues work on CoolProp#2245.

In saturation_PHSU_pure, we unspecify the phase of SatL and SatV to
perform a final check. However, for any future updates, these states
*must* be set with specified phase. This commit ensures that no matter
what happens (exception, etc.) the phase is always specified again.

This reverts commit c6b650b.

The commit caused potential recursive lookups and did not solve the
issue at hand.

The rhoV ancillary gave somewhat wrong results. This commit provides a
closer fit that prevents errors downstream.