Size is ignored when the output_bc_pattern implies it must be length 1, but shouldn't be ignored otherwise

Also handle core op caching key to support per-op invalidation

*Remove `structured_prefix` for Ops that map 0->0
*Do not introduce new structured ops that don't map 0->0 besides the ones that already existed

Internally Scan places those outputs to the right of nit_sot. Helper function reordering to match user definition was not handling this correctly

Co-authored-by: Ben Mares <services-git-throwaway1@tensorial.com>

Mark overly specific tests as xfail

Numba rewrites these to integers for specialized implementation

Also need to re-filter warnings because pytestmark.filterwarnings overrides library filters

When default `ignore_updates=True` for inc_subtensor, and boolean indices were rewritten during specialize

Also fix infinity identities for unsigned integers