[llvm] [RISCV] Combine vslidedown_vl with known VL and offset to a smaller LMUL (PR #66267)

[Philip Reames via llvm-commits]

https://github.com/preames requested changes to this pull request.

See previous batch of one off changes - sorry, got confused by the pr web interface.

To summarize:
* We need a more general solution to the vslidedown w/mask case.  Either via general select, or by removing the mask parameter on the node.  The current form is highly confusing.
* The vsrl using high lmul implies maybe we should go with the extract specific case as this lmul should be reducible as well.  

https://github.com/llvm/llvm-project/pull/66267