Extend DQ→MatMulNBits fusion to support Gemm + per-tensor/per-channel quantization#27769

jambayk merged 7 commits intomainmicrosoft/onnxruntime:mainfrom

jambayk/mnb-tensor-channel-gemmmicrosoft/onnxruntime:jambayk/mnb-tensor-channel-gemmCopy head branch name to clipboard