More consistent allocation of variables #68

I have come across the following issue that I may have introduced myself. Maybe (or most likely) it is because a lack of understanding on my side. So at the moment I believe what we make EBFM do is that it takes the ELMER mesh and assumes that each of the gridpoints needs a value for temperature and surface mass balance. I am fine with that assumption, because we could in theory build a glacier where we currently have no ice. However, I believe that EBFM also needs an ice mask which is a mask that delineates where we actually have ice surfaces. I thought that the variable grid["mask"] is used for that, but whenever I have fewer points in the array grid["mask"] then total gridpoints, there are a bunch of dimension mismatches. So my hunch is that we allocate certain variables based on the number of total gridpoints and certain variables on the number of ice-covered gridpoints. I think what should be done is that all variables are allocated based on the number of total gridpoints and then the approporiate ice-covered gridpoints are filled/marked accordingly.

To make this more concrete, at the moment we use the variable gpsum to allocate variables such as OUT["subW"]. But the variable gpsum is the number of glacier cells. In my view, this can be quite confusing. More critically, the model would fail immediately if gpsum < total number of gridpoints, for example in these lines here: https://github.com/EBFMorg/EBFM/blob/ElmerXIOSrun-patch-1/src/ebfm/INIT.py#L443C1-L450C74, because mask_indices[0] contains indices that go beyond the length of OUT["subW"]. Maybe @wardvp know how best to do this or if I am misunderstanding something...