transformers

1b33217a - Drop gguf_rename factory; use plain WeightRenaming/WeightConverter

Drop gguf_rename factory; use plain WeightRenaming/WeightConverter Follow repo conventions (mistral4 converter, Fp8Quantizer): the static GGUF→HF rule tables are now plain `WeightRenaming` chains plus `WeightConverter` for the few entries that need transform ops. `GGUFDequantize` is injected at the head of every WeightConverter by `GGUFQuantizer.update_weight_conversions` — same pattern as `Fp8Dequantize` / `Fp8Quantizer`.