remove redundant family table

update benchmark table: add decision-model-large (DI 58.52)

decision-model-large: Qwen3.8-Flash-Next-FP8 176B-A6B + merged rank-32 LoRA + jev pointer head (DI 58.52)

initial commit