Add regression test for C++ Delta Eddington solver

[mattldawson]

Adds a test of the new stub C++ Delta Eddington solver against the existing Fortran-based solver using the TUV 5.4 example conditions.

Notes:

• there is a check in the C++ test code for input conditions that are hard-coded to what is expected based on the configuration data. This is just to make sure the input data from Fortran are correctly mapped into the C++ data structures. If we decide to change the conditions this test runs under, this check will need to be modified
• the output of the stub solver function is set to predictable values, just to ensure the data transfer from C++ to Fortran is mapped correctly in the test. These can be replaced with the actual calculated radiation field, once they're available

closes #66

[sjsprecious]

@mattldawson I ran your latest branch on Derecho with GNU compiler and got the following failed tests:

         73 - regression_photolysis_rates (Failed)
         74 - regression_dose_rates (Failed)
         75 - regression_aerosols_radiator (Failed)
         76 - regression_aerosols_radiator_4strm (Failed)
         77 - regression_o2_radiator (Failed)
         78 - regression_o2_radiator_4strm (Failed)
         79 - regression_o3_radiator (Failed)
         80 - regression_o3_radiator_4strm (Failed)
         81 - regression_rayleigh_radiator (Failed)
         82 - regression_rayleigh_radiator_4strm (Failed)
         83 - regression_all_radiators (Failed)
         84 - regression_all_radiators_4strm (Failed)

Are these failures expected?

[sjsprecious]

egression_photolysis_rates

According to Matt, all the failed regression tests are due to the missing of loading conda module on Derecho, which we won't worry about it at this moment.