Rewrite to rely on `CFDataset` objects instead of `spatRaster`

[ErikKusch]

We are moving to ncdfCF CFDataset objects for their better support of metadata handling and more flexibility. This will require a sizeable rework of the existing codebase.

[pvanlaake]

There are likely going to be areas where the ncdfCF package is going to need some work to better support ClimHub functionality. Just let me know where you run into trouble and I'll look into it. The corollary to this is that ClimHub should import the latest dev version of ncdfCF for the foreseeable future. Currently, that version is 0.8.0.9000. Version 0.8.0 was released just 4 days ago but there is already some fresh functionality. Try this for instance:

library(ggplot2)
library(ncdfCF)
library(ClimHub)

NORA3 <- Access_NORA3(variable = "TS", dateStart = "2001-08-01 00", dateStop = "2001-08-01 21", 
                      extent = c(0, 30, 50, 75), leadTimeHour = 3)
ts <- NORA3[["TS"]]
ggplot() + geom_ncdf(data = ts$subset(time = "2001-08-01T15:00:00")) + coord_equal() + scale_fill_viridis_c()

Areas where I see current ncdfCF support:

• NC Write Operations #6
• Metrics_ETCCDI #14 and Metrics_BioClim #15
• Temporal_Aggregate #16

Areas for ncdfCF development:

• SpatialAggregate() with a factor argument

Not all at the same time of course...

[ErikKusch]

Amazing to see you progressed that! Guess, I have to relaunch #24 to remove the terra dependency there