Fix PALC Continuation Variable Boundary Targeting

[GrantHecht]

Currently, we employ a natural continuation-based method to step exactly to the upper or lower boundary for the continuation variable if a PALC step crosses one of the boundaries. The current approach has generally worked well, but recently, I've noticed it can occasionally get stuck, resulting in infinite reapplication of the natural continuation-based method. We need to fix this issue.

From initial observations, it seems that a bug exists that results in the acceptance of a step that crosses one of the bounds (such steps should not be accepted). When this occurs, the method then seems to (at least in some cases) get stuck repeatedly attempting and failing to repeatedly step exactly to the boundary.