Remove ZMM_RECOMMENDED_THRESHOLD workaround

[EgorBo]

This was a weird-looking workaround for the avx-512 throttle problem, since it is no more - I delete the workaround. Expecting a few diffs in aspnet collection, e.g.:

bool Test(ReadOnlySpan<byte> a, Span<byte> b) => 
    a.Slice(0, 100).TryCopyTo(b);

Codegen:

; Method Program:Test
       sub      rsp, 40
       vzeroupper 
       cmp      dword ptr [rdx+08H], 100
       jb       SHORT G_M62689_IG06
       mov      rax, bword ptr [rdx]
       mov      rdx, bword ptr [r8]
       mov      ecx, dword ptr [r8+08H]
       xor      r8d, r8d
       cmp      ecx, 100
       jb       SHORT G_M62689_IG04
-      vmovdqu  ymm0, ymmword ptr [rax]
-      vmovdqu  ymm1, ymmword ptr [rax+20H]
-      vmovdqu  ymm2, ymmword ptr [rax+40H]
-      vmovdqu  xmm3, xmmword ptr [rax+54H]
-      vmovdqu  ymmword ptr [rdx], ymm0
-      vmovdqu  ymmword ptr [rdx+20H], ymm1
-      vmovdqu  ymmword ptr [rdx+40H], ymm2
-      vmovdqu  xmmword ptr [rdx+54H], xmm3
+      vmovdqu32 zmm0, zmmword ptr [rax]
+      vmovdqu32 zmm1, zmmword ptr [rax+24H]
+      vmovdqu32 zmmword ptr [rdx], zmm0
+      vmovdqu32 zmmword ptr [rdx+24H], zmm1
       mov      r8d, 1
G_M62689_IG04:  
       mov      eax, r8d
       add      rsp, 40
       ret      
G_M62689_IG06: 
       call     [System.ThrowHelper:ThrowArgumentOutOfRangeException()]
       int3     
; Total bytes of code: 90

[ghost]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This was a weird-looking workaround for the avx-512 throttle problem, since it is no more - I delete the workaround (expect a few diffs)

Author:	EgorBo
Assignees:	EgorBo
Labels:	area-CodeGen-coreclr
Milestone:	-

[EgorBo]

A couple of diffs in aspnet collection (https://dev.azure.com/dnceng-public/public/_build/results?buildId=281207&view=ms.vss-build-web.run-extensions-tab) - it'd nice if we could have larger SPMI coverage for AVX-512.

A few size regressions becuase vxorps zmm, zmm, zmm is 6 bytes (evex) compared to 4 bytes for VEX. But the good news that we can always use the VEX form for it, same what native compilers do: