Merge pull request #2371 from RosettaCommons/aleaverfay/cart_min_deriv_unit_tests
This checkin fixes two bugs with Cartesian minimization of symmetric poses.
One was related to the recent change of symmetric atom tree minimizer to better support lattice symmetries. This change modified the edge weights on minimization graph, however CartesianMin still needed the old weights. This checkin changes the logic so the weights are modified as needed in SymAtomTree minimize, while keeping the minimization graph as it was before this change.
This checkin also fixes a longstanding bug in symmetric CartesianMinimization when only part of the structure only was allowed to move. In these cases, the domain map was built assuming an asymmetric pose, leading to not all energy edges getting calculated during cartesian minimization. This has been fixed (and @aleaverfay has added an integration test ensuring this works). Thanks Andrew for finding these issues.
Integration changes expected in all tests that minimize with symmetry.