vectorize-pr

[ipdemes]

This PR adds support for np.vectorize to cuNumeric
It depends on nv-legate/legate#640

This implementation has following limitations:


1. It requires all output arrays or scalars be the same type and shape. This is something that we will probably have to require, at least I don’t see the way to support this in a performant way

2. It requres all input arrays be the same shape. This is something we can fix with broadcasting.

3. There is currently no support for “excluded” and “signature”



I had to manually add typings for Numba to avoid pre-commit errors. I don’t think I did it right and asked @bryevdv to look at it. 







And thank you @bryevdv for helping with fixing documentation issues