[LongVec] Add Dot Product test cases

[farzonl]

resolves #1443

In Clang we have 3 unique dot product intrinsics
covering floating point, unsigned, and signed integers.

These test cover these 3 unique paths.

[farzonl]

The intel failures are XPasses. There is one legit failure on Metal that I will need to investigate before asking for reviews:

******************** TEST 'OffloadTest-mtl :: Feature/HLSLLib/dot.longvec.fp32.test' FAILED ********************
Exit Code: 1

Command Output (stdout):
--
# RUN: at line 103
/Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/bin/split-file /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/OffloadTest/test/Feature/HLSLLib/dot.longvec.fp32.test /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp
# executed command: /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/bin/split-file /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/OffloadTest/test/Feature/HLSLLib/dot.longvec.fp32.test /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp
# RUN: at line 104
"/Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/DXC/build/bin/dxc-3.7" -T cs_6_9 -Fo /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp.o /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp/source.hlsl
# executed command: /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/DXC/build/bin/dxc-3.7 -T cs_6_9 -Fo /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp.o /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp/source.hlsl
# RUN: at line 105
/Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/bin/offloader -debug-layer /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp/pipeline.yaml /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp.o
# executed command: /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/bin/offloader -debug-layer /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp/pipeline.yaml /Users/admin/actions-runner/_work/offload-test-suite/offload-test-suite/llvm-project/build/tools/OffloadTest/test/mtl/Feature/HLSLLib/Output/dot.longvec.fp32.test.tmp.o
# .---command stdout------------
# | Using Metal API
# | Descriptor heap created with 3 descriptors.
# | Creating SRV: { Size = 116, Register = t0, Space = 0 }
# | Creating SRV: { Size = 116, Register = t1, Space = 0 }
# | Creating UAV: { Size = 12, Register = u2, Space = 0, HasCounter = 0 }
# | SRV: HeapIdx = 0 EltSize = 4 NumElts = 29
# | SRV: HeapIdx = 1 EltSize = 4 NumElts = 29
# | UAV: HeapIdx = 2 EltSize = 4 NumElts = 3
# | DescriptorRange[0] { Type=0, NumDescriptors=1, BaseShaderRegister=0, RegisterSpace=0, OffsetInDescriptorsFromTableStart=0 }
# | DescriptorRange[1] { Type=0, NumDescriptors=1, BaseShaderRegister=1, RegisterSpace=0, OffsetInDescriptorsFromTableStart=1 }
# | DescriptorRange[2] { Type=1, NumDescriptors=1, BaseShaderRegister=2, RegisterSpace=0, OffsetInDescriptorsFromTableStart=2 }
# | Resource[0] { Type=Table, Space=4294967295, Slot=4294967295, TopLevelOffset=0, SizeInBytes=8 }
# `-----------------------------
# .---command stderr------------
# | Validation Warning: [ mvk-warn ]
# | VK_ERROR_FEATURE_NOT_PRESENT: Metal does not support buffer robustness.
# | 
# | gpu-exec: error: Failed to compile and link DXIL to Metal IR: IRErrorCodeUnsupportedInstruction
# `-----------------------------
# error: command failed with exit status: 1

I'm not sure why we are getting a VK_ERROR_FEATURE_NOT_PRESENT but it seems unrelated.

The real issue is IRErrorCodeUnsupportedInstruction Maybe the metal IR does not support vectorized dot product. I'm thinking the vectorized signed and unsigned integer dot products are working because they were always implemented as as a vector reduce and sumation, but floats lowered to special hardware units and never went through that kind of math before.