setting patch%itype to invalid for FATES patches

[rgknox]

In #2932 we found a regression where FATES runs were allowed to erroneously perform operations with the patch%itype structure. I think we've allowed the allocation of this array in fates runs because of the anticipation of allowing non-fates modules (that would need access to itype) to run along side fates.

Setting values of patch%itype for fates patches back to the uninitialized value would help prevent issues that were experienced in #2932, because calling that array will result in out of bounds errors.

You can see we played with the idea of doing this here:

https://github.com/ESCOMP/CTSM/blob/ctsm5.3.018/src/utils/clmfates_interfaceMod.F90#L933

[ekluzek]

I love this idea @rgknox. This will help us find problems when cases are run in DEBUG mode. I'm hoping we could bring this in soonish, but I'm also afraid that this will expose existing problems that might take some time to work through. But, that would also be a benefit of doing this. I've assigned you and @glemieux to it -- but please change as appropriate. I also put it as something to do by the answer changing freeze, but hope this can come in ASAP.

Since, this is JUST changing the value over FATES columns what you have here should be fine for working towards our desire to have both FATES and non-FATES columns running at the same time. So I don't think your second sentence applies for what you've proposed.

[rgknox]

Created this PR to address itype, just a strawman but does the job: #2935
Should we combine this with #2934 ?

[wwieder]

Given the urgency of this PR, @rgknox are you OK with the suggested order on upcoming tags?