Room for optimization in stump using unsigned integer indices

[alvii147]

trafficstars

This is in reference to the discussion around the differences in using np.int64 and np.uint64 to index numpy errors in this PR and this numba discussion thread.

Based on the prior discussions and experiments, it may be possible to optimize the stump module (and potentially other modules), specifically by replacing index values of data types np.int64 to np.uint64.

For example, the iter_range variable here may be replaced with an np.arange call using dtype=np.uint64.