[llvm] [LLVM] Add `llvm.masked.compress` intrinsic (PR #92289)

[Lawrence Benson via llvm-commits]

================
@@ -24975,6 +24975,72 @@ The '``llvm.masked.compressstore``' intrinsic is designed for compressing data i
 
 Other targets may support this intrinsic differently, for example, by lowering it into a sequence of branches that guard scalar store operations.
 
+Masked Vector Compress Intrinsic
----------------
lawben wrote:

The more I think about it, the more I think you are right about the naming. The "mask" is an essential part of the operation, as @nikolaypanchenko points out, so it does not make sense to use it to distinguish between a masked and non-masked compress. 

Pulling in @efriedma-quic @RKSimon for an opinion on naming and putting it in the experimental namespace.

If I change it, I'd probably change it to `@llvm.experimental.vector.compress` as suggested by @lukel97 , as this covers both the `vector` and `experimental` (see [other comment](https://github.com/llvm/llvm-project/pull/92289/#discussion_r1646842999) for discussion on this) for now.

https://github.com/llvm/llvm-project/pull/92289