`model2netcdf.ED2()` calculations for -E- files are incorrect

[Aariq]

Bug Description

When I re-wrote code in read_E_files(), I didn't have a full understanding of the .h5 files output by ED2 (also the metadata in those files was wrong). The PFT-level monthly variables output in the .nc files should not be trusted. I know how to do the calculations correctly now (I'm fairly certain), so I'll make a PR shortly.

Screenshots

Here's some comparisons of data calculated (correctly) from the .h5 files vs. what's in the .nc files.

[Aariq]

Calculations done correctly (I hope) here: https://github.com/cct-datascience/ed2-mandifore/blob/main/R/extract_E_files.R

[KristinaRiemer]

Can this be moved to closed?