Feature/jkolassa/gndtmp cn cleanup

[gmao-jkolassa]

This PR addresses issue #530 and involves a cleanup of subroutines gndtmp and gdntmp_cn.

Subroutine gndtmp was modified to make several input arguments optional and make parts of the code dependent on the presence of these optional input arguments. As a result gndtmp can now be called instead of gndtmp_cn while still retaining all previous functionality. The now obsolete subroutine gndtmp_cn was removed.

Two points to note/discuss:

1. I had to change the order of the input arguments in gndtmp in order to avoid having to make all inputs keyword arguments. I have updated all calls to gndtmp accordingly. If the preference is that the input order remains the same, I can make the necessary changes, but it will mean that all input arguments have to be passed as keyword arguments.
2. This branch has been created from 'develop' and as a result it does not contain the changes to gndtmp(_cn) and the temperature variable names that have been implemented as part of the peat PR. Depending on when each PR is expected to be merged, I can include the changes that were made in the peat PR here.

I have tested this new branch in 3 one-month offline experiments using Catchment, Catchment-CN4.0 and Catchment-CN4.5. For all three experiments the model outputs after one month were zero-diff with respect to a simulation generated with 'develop'. I have not run any coupled simulations.

[gmao-rreichle]

preliminary 0-diff tests successful:

• AGCM (Biljana)
• GEOSldas (Rolf) [as expected, tests with aggressive build had roundoff differences]