Model updates: eCLM with new snow thermal conductivity options; ParFlow with more CUDA13 fixes

[kvrigor]

Model updates containing these fixes:

• eCLM: Snow thermal conductivity changes: Sturm1997 and Jordan1991
• ParFlow: Applied cudaMemPrefetchAsync API change for CUDA 13

Please note: eCLM beta-0.6 is a breaking change! TSMP2 sims have to update the lnd_in namelist with these new required parameters below. For more info see linked PR above.

&clm_inparm
 snow_thermal_cond_method = 'Sturm1997'
 snow_thermal_cond_glc_method = 'Sturm1997'
 snow_thermal_cond_lake_method = 'Jordan1991'

[jjokella]

From yesterday's meeting, I gather that still a test of the EUR-11 case is pending, but from my side PR is approved.

[kvrigor]

@s7yoewer already did a quick test in this eCLM issue:

While I get some differences in simulated surface SWC, the main issue persists: also with the new config, surface SWC values above 1 are still visible for many parts of Europe in frozen conditions.

The Sturm1997 parameterization doesn't solve Yorck's issue; however this does not necessarily invalidate this new snow parameterization scheme. What's important is that the old model behavior, i.e. snow_thermal_cond_method = 'Jordan1991' is still available and is the default setting in the lnd_in eCLM namelist.