Feature/uxgrid morton hashing

[fluidnumerics-joe]

No description provided.

[fluidnumerics-joe]

One concern with this PR at the moment is that, when run under pytest, the test_fesom2_square_delaunay_antimeridian_eval gets killed on my system with an OOM issue. This is related to @erikvansebille 's concerns about high memory usage in the morton hash table construction .

[fluidnumerics-joe]

In the generic unstructured antimeridian sample dataset, I did find some bad elements (They have a single longitude value but span the whole range of latitude). Looking to see if this dataset needs some correction.

[fluidnumerics-joe]

I was able to track down the bad elements in the antimeridian test case that were causing an insanely large number of hash table entries ( O(10^6) ). The boundary mask needed to be corrected. I also adjusted this _fesom2_square_delaunay_antimeridian domain boundary to be a bit smaller to keep the memory footprint manageable for unit testing. This one should be good to go, pending a review.