Refactor Llama generator, helpers, and build assets

make Gemma 4 the default model, enable thinking mode
style updates

pipeline/includes/data_model/application_options.h

@@ -0,0 +1,42 @@
+#ifndef BIERGARTEN_PIPELINE_INCLUDES_DATA_MODEL_APPLICATION_OPTIONS_H_
+#define BIERGARTEN_PIPELINE_INCLUDES_DATA_MODEL_APPLICATION_OPTIONS_H_
+
+/**
+ * @file data_model/application_options.h
+ * @brief Program options for the Biergarten pipeline application.
+ */
+
+#include <cstdint>
+#include <string>
+
+/**
+ * @brief Program options for the Biergarten pipeline application.
+ */
+struct ApplicationOptions {
+   /// @brief Path to the LLM model file (gguf format); mutually exclusive with
+   /// use_mocked.
+   std::string model_path;
+
+   /// @brief Use mocked generator instead of LLM; mutually exclusive with
+   /// model_path.
+   bool use_mocked = false;
+
+   /// @brief LLM sampling temperature (0.0 to 1.0, higher = more random).
+   float temperature = 1.0F;
+
+   /// @brief LLM nucleus sampling top-p parameter (0.0 to 1.0, higher = more
+   /// random).
+   float top_p = 0.95F;
+
+   /// @brief LLM top-k sampling parameter.
+   uint32_t top_k = 64;
+
+   /// @brief Context window size (tokens) for LLM inference. Higher values
+   /// support longer prompts but use more memory.
+   uint32_t n_ctx = 8192;
+
+   /// @brief Random seed for sampling (-1 for random, otherwise non-negative).
+   int seed = -1;
+};
+
+#endif  // BIERGARTEN_PIPELINE_INCLUDES_DATA_MODEL_APPLICATION_OPTIONS_H_