[AMD] Enable fused shared expert append and flatten quant for fp8 deepseekR1 model#13705

HaiShaw merged 4 commits intosgl-project:mainsgl-project/sglang:mainfrom

yichiche:fused_append_shared_expertsyichiche/sglang:fused_append_shared_expertsCopy head branch name to clipboard