llama: end-to-end tests (#19802)

* tests: add end-to-end tests per model architecture

* fixup for rebase

* fix use-after-free in llama-model-loader.cpp

* fix CI

* fix WebGPU

* fix CI

* disable CI for macOS-latest-cmake-arm64

* use expert_weights_scale only if != 0.0f

* comments

src/llama-vocab.cpp

@@ -1719,7 +1719,7 @@ private:
 };
 
 void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
-    struct gguf_context * ctx = ml.meta.get();
+    struct gguf_context * ctx = ml.metadata;
 
     // determine vocab type
     {