[llvm] [X86][AVX] Fix handling of out-of-bounds shift amounts in AVX2 vector logical shift nodes #83840 (PR #86922)

[Simon Pilgrim via llvm-commits]

https://github.com/RKSimon commented:

Sorry - this fell off my radar. One last request and this is ready to go - we can iterate later.

https://github.com/llvm/llvm-project/pull/86922