Amazon precip bias in coupled simulations

[kjetilaas]

In the last couple of NorESM development meetings, we have been coming back to the dry precipitation bias in NorESM2.5_x simulations. See for instance the figure below from NorESM2.5_alpha08d coupled simulation:
NorESMhub/noresm3_dev_simulations#67

This issue is also found in current CESM3 b simulations (but not in f cases):
NCAR/cesm_dev#26

We also see much more Amazon precip in f cases, like:
NorESMhub/noresm3_dev_simulations#71

See comparison here:

[kjetilaas]

It is worth noting here that NorESM simulations here are with FATES-SP. That means:

• different vegetation component from CESM
• much more constrained vegetation response (since LAI is prescribed)

@oyvindseland @rosiealice