RRTMGP refactoring

[dustinswales]

This PR contains changes to the RRTMGP physics.

• Major refactoring of RRTMGP interstitials. collapse all rrtmgp_lw(sw) scheme files into rrtmgp_lw(sw)_main schemes. This reduces the number of Interstitial fields needed by the GP scheme and allows us to use the rte-rrtmgp submodule directly, w/o the need for a special branch containing the CCPP hooks.
• For longwave radiation calculation, add "blocking" capability to main loop.
• Fix bug in coupling of RRTMGP shortwave radiation to surface.
• CMake cleanup.

[grantfirl]

@dustinswales I'm finished looking through this. Looks fine to me for the most part, with some worries about uninitialized intent(out) variables. It seems like every small scheme that was deleted ended up in the SW or LW main schemes? It's a little difficult to follow code moving between files in the GitHub interface (without looking really hard).

[dustinswales]

@dustinswales I'm finished looking through this. Looks fine to me for the most part, with some worries about uninitialized intent(out) variables. It seems like every small scheme that was deleted ended up in the SW or LW main schemes? It's a little difficult to follow code moving between files in the GitHub interface (without looking really hard).

@grantfirl Thanks for your review. I addressed your concerns about the intent of some variables.

I admit this is hard to follow, my apologizes, it's a total restructuring of how GP is implemented in the CCPP.

[jkbk2004]

We are about to start working on ufs-community/ufs-weather-model#1571. @dustinswales can you mark out unsolved conversations?