SP restart crash

[rgknox]

We do some reading/writing of restart variables using a loop that spans:
numpft x nlevsclass x n_term_mort_types

In non SP runs, we most likely allocate enough memory because we typically run with more cohorts per gridcell than the product of these three terms, but in SP mode that is not likely true.

Here is an example of traversing that loop.

https://github.com/NGEET/fates/blob/sci.1.72.6_api.34.0.0/main/FatesRestartInterfaceMod.F90#L2199-L2217

I believe the solution is to increase the allocation here by adding in n_term_mort_types to the product with nlevsclass and numpft:

https://github.com/NGEET/fates/blob/sci.1.72.6_api.34.0.0/main/FatesInterfaceMod.F90#L879-L880

to be:

fates_maxElementsPerSite = max(fates_maxPatchesPerSite * fates_maxElementsPerPatch, &
              numWatermem*numpft, num_vegtemp_mem, num_elements, nlevsclass*numpft*n_term_mort_types)

I'm testing now.

[rgknox]

Noting that after some discussion, @ckoven thought it would be better to not report these diagnostics at all, since they are irrelevant with SP