[REQ] Support for Python slicing syntax

[mmacklin]

Description

Users should be able to select sub-ranges of arrays and tiles inside kernels, e.g.:

t = wp.tile_load(arr[i:i+m, j:j+n])
t = wp.tile_load(arr[i:i+m])

Context

More flexible indexing into arrays would allow writing e.g.: segmented scan with Warp tile kernels.