[Bug tree-optimization/93131] ((a&8) == 8) && ((a&2) == 2) is not optimized to (a&(8|2)) == (8|2)

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93131

--- Comment #14 from Richard Biener <rguenth at gcc dot gnu.org> ---
Don't we want this in reassoc where I believe we do related transforms?  That's
IMHO better than recursively pickling large expressions via match.pd patterns
(which incidentially has a recursion limit)