ReconnectBoundaryB3 does not conserve energy

[bprather]

Perhaps part of the reason that setting boundaries/reconnect_B3_XX was so calming to simulations was that it was leaking magnetic energy.

Basically, the reconnection function reduces the local magnetic pressure by reducing the B3 magnetic field component, but does not then update the fluid primitive variables to move that energy into the fluid internal energy. Since the primitive variables are the next values to be used, the total energy is effectively reduced, disappearing the magnetic field's energy into nothing instead of shifting it to the fluid.

This can be fixed by simply calling U->P on affected zones to update the fluid primitive variables in light of the new magnetic field, PR coming soon. Thanks to @hyerincho for finding this!

Note this doesn't affect CancelBoundaryU3 and CancelBoundaryT3 -- though, they had stability issues of their own addressed in #126