Fix IC staging bug: should be idatestr not rdatestr

[CoryMartin-NOAA]

          I think this was supposed to be `idatestr` not `rdatestr`. If you create cold start ICs from ufs_utils, there won't be a rdate directory.

Originally posted by @CoryMartin-NOAA in #1529 (comment)

[DavidHuber-NOAA]

For warm starts, atm data is linked from the previous cycle, so you would want rdatestr for those. For cold starts, we should be looking at idatestr. I have a branch with this fix in place. So far I have only tested it for cold starts, but I will test it for warm starts as well before opening a PR.

[DavidHuber-NOAA]

While testing this for warm starts, I encountered errors for missing increment files (ensemble ratmi*.nc and deterministic atmi*.nc and atminc.nc). Reading through the setup_expt.py script, only a handful of files are linked from the source ICs' analysis/atmos/ directories (abias, abias_pc, abias_air, and radstat), but additional files are needed to support IAU.

@WalterKolczynski-NOAA Is there a reason I would not want to link in the entirety of the analysis/atmos/ directory from the warm start ICs (i.e. calling link_files_from_src_to_dst)?

[WalterKolczynski-NOAA]

I think we'd rather have an explicit list. That way you only copy what you need and also know if something is missing. @aerorahul?

[aerorahul]

There are 3 ways to start the atmosphere:

1. cold start
2. warm start w/o IAU
3. warm start w/ IAU

1 is the way most (99%) of developers start their experiment (generate cold start ICs offline from ops).

2 is also functional if someone wants to start from warm start ICs but no IAU.

For 3; I think there are blockers in the forecast scripts that will prevent doing this; i.e. IAU is turned off for the first half cycle (I will need to check to be sure)

Is 3 a mode we want to support? Yes. Is it high on the priority list? Probably not worth the effort ATM.

[DavidHuber-NOAA]

@aerorahul Noted, thanks. I will continue my warm-start testing with IAU turned off then.