the attached two patches remove TOV_Atmosphere from the code and do not set the conservative variables. Instead TOVSolver relies on GRHydro to set the conservatives in InitialPrim2Con and to enforce atmosphere in initialatmospherereset. This also fixes any asymmetry in atmosphere handling between the stars.
Most tests pass without change (since GRHydro ran anyway), the failing one being the test_two_av TOVSolver test which the second one regenerates. Differences are small but above threshold in the average case.