Fix and restrict NVFP4 edge-cases in llama-graph (#24331)

* Move post-GEMM MUL required for dequant b4 lora and bias add

see https://github.com/ggml-org/llama.cpp/pull/23484 :
1. For lora, I would presume we want fully dequantized values before
   doing the residuals, but this depends on how the LORAs were
generated. Literature tells me LORA happens post-mul but pre-bias add https://github.com/ggml-org/llama.cpp/pull/8332
2. For ModelOPT, bias-add should happen on [fully-dequantized
   values](https://github.com/NVIDIA/Model-Optimizer/blob/b49f9b9e2d747af992d78a3aa7f10efe5a8847e1/modelopt/torch/quantization/backends/nvfp4_gemm.py#L59-L64)

* Restrict build_ffn for NVFP4 to supported combinations

src/llama-graph.h

@@ -853,11 +853,12 @@ struct llm_graph_context {
               ggml_tensor * cur,
               ggml_tensor * w_s = nullptr) const;
 
-    // do mat_mul_id, while optionally apply lora
+    // do mat_mul_id, while optionally apply lora and per-expert scale
     ggml_tensor * build_lora_mm_id(
               ggml_tensor * w,   // ggml_tensor * as
               ggml_tensor * cur, // ggml_tensor * b
-              ggml_tensor * ids) const;
+              ggml_tensor * ids,
+              ggml_tensor * w_s = nullptr) const;
 
     ggml_tensor * build_norm(
              ggml_tensor * cur,