Changes

ADG1_w_closure consistency fix (#1282)

* G_unit_test / ADG1_w_closure fix. A recent commit (bff619bdb2836f83ac658be8b8c258d41aaa6e14) introduced a change to ADG1_ADG2_responder_params that caused an inconsistency in the pdf calculations that the G_unit tests picked up on. This corrects it, but is also BIT_CHANGING, since it changes the calculation slightly when wp2 is exactly tolerance.

* Adding test to ensure err_info is correctly set by adg1 routines when calling with bad inputs - also moving wp2 check to ADG1_w_closure because it makes more sense there

Fixing clubb_error_handling_gfortran_test (#1281)

* Fixing error test by making the code check for NaNs in all solver cases - before it was only checking in Lapack. Also making a python version of the test and updating jenkins script to use it.

* A little cleanup

* Adding more comments and making a little data structure for the tests