R21C HISTORY: model/corrected precip and runoff

A couple of things related to R21C file specs did not get done in #637:

1. Rolf: Figure out how to deal with LONG_NAME of model vs. corrected precipitation. In the MERRA-2 file specs, the words "bias corrected" were appended somehow (manually?) to LONG_NAME for PRECTOTLAND and PRECSNOLAND.
   a. "Bias corrected" is not a good term; going forward, we should use "observation corrected" because the corrections are applied at the daily scale.
   b. Adding "observation corrected" to the LONG_NAME in *GridComp.F90 files requires thought because currently the F90 file supports both uncorrected and corrected precip over land in the same export variable.
   See original comment.

   UPDATED 31 July 2024 by @gmao-rreichle:
   The LONG_NAME for the land-average precip exported from SurfaceGC is now "Total_precipitation_land", which does not allude to the presence or absence of the observation-based correction. I'm still unsure if it's possible to add some language about "corrected" into the LONG_NAME depending on whether obs precip files are used or not. Since this would have to be done dynamically by detecting the use of obs precip files, and since HISTORY specs are something that is presumably done during initialization, I doubt this would work. (Note that there should only be one Surface GC F90 file; we do not want to make this custom for each product.) For now, I think it's ok to leave as is, which also means that we do not need to finesse the language about "bias-corrected" vs. "observation-corrected". Note that the SHORT_NAME is customized for each product in the product-specific HISTORY.rc file. This was done for M21C in https://github.com/GEOS-ESM/GEOSgcm_App/pull/621

2. Lauren: 'RUNOFF' exported from 'SURFACE' in the flx collection should provide combined runoff from land and landice. However, a test indicated that the entire grid are no data values. 'RUNOFF' from LANDICE exports correctly, so the details of how this export propagates up through to SURFACE from LAND, LANDICE, & LAKE are beyond me.
   See original comment.

3. The long name changes in #637 need to find their way into the GCM GC develop branch so they will eventually find their way into all GMAO systems.

cc: @lcandre2 @sdrabenh @elakkraoui @lltakacs @rlucches

Start by tracing the RUNOFF export from SURFACE through LAND, LANDICE, and LAKE, using the SurfaceGC and GridComp.F90 references in the issue. Review the HISTORY metadata changes from #637 and the product-specific HISTORY.rc handling. Done means determining the correct combined runoff output and documenting whether the requested long-name changes can be supported.