llama: support both embd + raw tokens in batch (#29622)

* llama: support both embd + raw tokens in batch

* add to test-llama-archs

* also check case llm_arch_supports_mixed_batch = false

* constant graph topology

* have dedicated input for mixed case

* rm set_tensor_backend

* is_embd --> type

* consolidate m-rope pos handling into one place

* nits

src/models/dflash.cpp

@@ -604,6 +604,7 @@ llama_model_dflash::graph<false>::graph(const llama_model & model, const llm_gra
     };
 
     // KV cache injection
+    ASSERT_EMBD_OR_TOKEN(ubatch);
     if (ubatch.embd) {
         auto inp = std::make_unique<llm_graph_input_embd>(n_embd_inp);
 
@@ -870,6 +871,7 @@ llama_model_dflash::graph_dsv4::graph_dsv4(const llama_model & model, const llm_
     llm_graph_input_attn_k_iswa * inp_attn = build_attn_inp_k_iswa();
 
     // KV cache injection: fused target features from the encoder
+    ASSERT_EMBD_OR_TOKEN(ubatch);
     if (ubatch.embd) {
         auto inp = std::make_unique<llm_graph_input_embd>(n_embd_inp);