algorithms: bounds

[IvanGrigorik]

Add lower and upper bound using binary search
Tested for CUDA and OpenMP

API similar to standard thrust::upper_bound: https://nvidia.github.io/cccl/thrust/api/group__binary__search_1gac85cc9ea00f4bdd8f80ad25fff16741d.html

# Thrust:
thrust::upper_bound(ForwardIterator first, ForwardIterator last, const LessThanComparable &value ) # -> return iterator
thrust::lower_bound(ForwardIterator first, ForwardIterator last, const GreaterThanComparable &value ) # -> return iterator

# Ours:
pk.upper_bound(view: pk.View, lastIndex: int, lessThanValue: int) # -> return int
pk.lower_bound(view: pk.View, lastIndex: int, greaterThanValue: int) # -> return int

Same for lower_bound
One of the student asked to implement this in order to use it from pykokkos workunits, since they needed this operation in their research

[gliga]

Is this closing any PR? Why is this necessary? Are we matching any API?
This should all be documented in the description of the PR.

[IvanGrigorik]

Updated description

[gliga]

Is it sufficient to work only with ints?

[IvanGrigorik]

For some reason tests fails due unknown reason. I couldn't reproduce but on my local machine nor on our local server
@kennykos was able to reproduce issue, but I cant help him, since it's on the server that I can not access