Add support for some more simd folding to improve the new IndexOf and related helpers

[tannergooding]

This handles cases like Vector128.IndexOfWhereAllBitsSet(Vector128.Equals(x, y)) and ensures we get more optimal codegen.

; Method ClassLibrary2.Class1:M(System.Runtime.Intrinsics.Vector128`1[int],System.Runtime.Intrinsics.Vector128`1[int]):int (FullOpts)
G_M000_IG01:                ;; offset=0x0000

G_M000_IG02:                ;; offset=0x0000
       C5F81001             vmovups  xmm0, xmmword ptr [rcx]
       C5F97602             vpcmpeqd xmm0, xmm0, xmmword ptr [rdx]
-      C5F176C9             vpcmpeqd xmm1, xmm1, xmm1
-      C5F176C0             vpcmpeqd xmm0, xmm1, xmm0
       C5F850C0             vmovmskps rax, xmm0
       F30FBCC0             tzcnt    eax, eax
       B9FFFFFFFF           mov      ecx, -1
       83F820               cmp      eax, 32
       0F44C1               cmove    eax, ecx

G_M000_IG03:                ;; offset=0x0023
       C3                   ret      
; Total bytes of code: 36

This mostly stems from ensuring that Compare(mask, AllBitsSet) and Compare(mask, Zero) are handled. However, it also found some cases where we weren't folding TrailingZeroCount, ReverseElementBits, or PopCount which are used by the new helpers and so ensures those are handled as well.

[tannergooding]

CC. @dotnet/jit-contrib

Fixes some codegen issues that @EgorBo saw. Going to have a follow up PR that better improves the Arm64 handling by ensuring that ExtractMostSignificantBits is carried through to lowering so it can be properly optimized based on these types of transforms (giving it the same lightup as x64)

[EgorBo]

/azp run Fuzzlyn

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[tannergooding]

Fuzzlyn issue is #117605

[tannergooding]

/ba-g unrelated android timeouts