[llvm] [AVX-512] make vpternlogq more aggressive for longer chains of bitmanipulations (PR #189971)

[Julian Pokrovsky via llvm-commits]

raventid wrote:

@mightsleep we had an idea, that current implementation is just a lightweight squasher of ternlogs and we could do a more aggressive optimisations in the middle-end (or backend). But I don't have a finished design to share!

Your patch looks cool 👍 


https://github.com/llvm/llvm-project/pull/189971