Evaluate parallelization of Add_snow_increment in GDASApp

[tsga]

For 2DVar increments, the current v17 implementation uses a single increment and adds it to ensemble (80) members by looping through each member. A modification to the land_inc_add and pygfs codes enables adding increments in parallel.

[tsga]

Changes were made in GDASApp
https://github.com/tsga/GDASApp/tree/snowda_deterinc_ensrest
and global workflow
https://github.com/tsga/global-workflow/tree/snowda_deterinc_ensres

Tests in WCOSS for snowens_anl with 80 ensemble members of C384 revealed that the add increment section of the task only take ~10sec of 21 minutes that the enkfgdas_snowensanal takes (with ~12 minutes spent on copying of files, 6 minutes for GDAS/JEDI execution, and ~1min obs processing).

Therefore parallelizing only add increment is not worth it.

Shelfing this PR, possibly to revive it once approach to efficient data copy is identified.