Add Q6_0_ROCMFPX_AGENT+imatrix (~31GB)

Add sealed Q4_0_ROCMFP4_COHERENT+imatrix (~19GB) — preferred AMD-calibrated quant

Add imatrix used for COHERENT sealed quant

Add Qwen3.6-35B-A3B-Q4_0_ROCMFP4.gguf (~22GB single-file ROCmFPX quant)

initial commit