[llvm] [RISCV] Use LMUL=1 for vmv_s_x_vl with non-undef passthru (PR #66659)

[Craig Topper via llvm-commits]

topperc wrote:

> > Have we arrived at a point where we just never reach isel with a VMV_S_X with a register class other than LMUL=1?
> 
> I think so. Any clean up you see resulting from that? Are there some pseudos we could delete, or are those still required by mca?

Should we create it with an LMUL=1 type with surrounding subvector ops in the first place instead of fixing it with a DAG combine?

Does mca use the pseudos? @michaelmaitland 

https://github.com/llvm/llvm-project/pull/66659