debug allskywithclear instability

[juliasloan25]

In the latest set of dependency updates, #598, our two shortruns using the allskywithclear radiation option have become unstable. For now, they've been changed to use gray radiation, but we should address the root cause of this and bring back allsky stability.

changed toml files to revert:

• config/model_configs/coarse_single_ft64_hourly_checkpoints_co2.yml
• config/model_configs/target_params_in_slab_test2.yml

cc @LenkaNovak @akshaysridhar

[LenkaNovak]

Follow up:

• local checks reverting TD to 11.2 still leads to instability after 2d 8h (slabplanet) - https://github.com/CliMA/Thermodynamics.jl/pull/185/files is not the culprit
• coupler and at mos config arguments pre post CA0.20 updates have not changed substantially
• rayleigh_sponge: true, implicit_diffusion: true and approximate_linear_solve_iters: 2, dt = 300s did not help

[LenkaNovak]

The allskywithclear is now stable for 200d at the 100km target resolution with all AMIP configurations. build

[juliasloan25]

@LenkaNovak should we change coarse_single_ft64_hourly_checkpoints_co2.yml and target_params_in_slab_test2.yml back to use allskywithclear radiation now that it's stable?

[LenkaNovak]

Yes, we can change these (or add new ones), but the important one was for allskywithclear work with the target run. Note that it is unclear whether allskywithclear should work with the lower resolutions, since atmos doesn't have comparable tests.