Logical & TruncConv Ops + Tests

[Krovatkin]

This PR includes

• logical operations + tests
• truncation & conversion operations for 32x4 types + tests

---

This change is 

[Krovatkin]

@Cellule @MikeHolman I'll ping you guys later about this PR when all the tests pass
fyi @arunetm

[Krovatkin]

@chakrabot Please test Ubuntu shared_ubuntu_linux_debug

[Krovatkin]

@Cellule Could you please take a look when you have a moment?

[MikeHolman]

I think nonstandard true values will be broken. Currently, if a lane is non-zero it is considered true. Lowering code uses PMOVMSKB, which only looks at the most significant bit of each byte.

https://github.com/WebAssembly/simd/blob/master/proposals/simd/SIMD.md#boolean-horizontal-reductions

[Krovatkin]

@MikeHolman absolutely! will fix!

[Krovatkin]

@chakrabot Please test Windows ci_slow_x64_debug

[Krovatkin]

@arunetm okay

[Krovatkin]

@MikeHolman Looks like we should be able to handle non-canonical boolean values now.

[Krovatkin]

@MikeHolman @Cellule Hey guys! Could you please take another look? I'd like to push this one and 2 more then I'll rebase wasm.simd as @Cellule suggested.

[Cellule]

Reviewed 9 of 11 files at r1, 8 of 8 files at r2.
Review status: all files reviewed at latest revision, 4 unresolved discussions.

---

lib/Backend/LowerMDShared.h, line 374 at r2 (raw file):

    IR::Instr*          Simd128LowerAnyTrue(IR::Instr* instr);
    IR::Instr*          Simd128LowerAllTrue(IR::Instr* instr);
#ifdef ENABLE_WASM

Could we start using a different #define for wasm.simd ? Just so we can more easily organize what's what.
Something like ENABLE_WASM_SIMD. I thin that will also help the transition to remove simd.js and keep what we need for wasm.

---

lib/Backend/LowerMDSharedSimd128.cpp, line 2273 at r2 (raw file):

    if (m_func->GetJITFunctionBody()->IsWasmFunction())
    {
        Js::OpCode spreadOpCode = Js::OpCode::InvalidOpCode;

nit: could we name this cmpOpcode to match how we use it

---

lib/Runtime/Language/SimdBool32x4Operation.cpp, line 45 at r2 (raw file):

    }

    template bool SIMDBool32x4Operation::OpAllTrue<int>(const SIMDValue& val);

nit: I would prefer use int32, int16, int8 instead of int, short, char

---

lib/Runtime/Language/SimdUtils.h, line 335 at r2 (raw file):

        {
#ifdef ENABLE_WASM
            T* cursor = (T*)val.i8;

I would add some CompileAssert here

            CompileAssert(sizeof(T) <= sizeof(SIMDValue));
            CompileAssert(sizeof(SIMDValue) % sizeof(T) == 0);
            T* cursor = (T*)val.i8;
            uint size = sizeof(SIMDValue) / sizeof(T);

---

Comments from Reviewable

[Cellule]

---

Reviewed 6 of 6 files at r3.
Review status: all files reviewed at latest revision, all discussions resolved.

---

Comments from Reviewable

[Krovatkin]

@arunetm Could you please merge this PR?

[Krovatkin]

@Cellule , Thanks a lot for your help!