AST-32: Hashing of Nat/Int

[ggreif]

No description provided.

[nomeata]

Is xoring the two halves really a good hash function for hash tables? Don't we want to approximation of a uniform distribution that spreads out across the whole range? (I am not an expert here, but it seems to be a bit navie to be true.)

[ggreif]

@nomeata This is the most naive hash function I could come up with without dropping bits. For starters, I wanted something simple that works for the interpreter and compilation.

I am open for algorithms with better uniform distribution properties (https://google.com?q=hash+function+for+64+bit+integer), but this should be enough, if we only want bucketisation (% p, for small prime p) for a hash table.

[nomeata]

Ok, good enough for getting this into master, can always change it later if we have to.