refactor(llm-api): implement clean architecture following scraper pattern

Split monolithic 1012-line main.rs into layered hexagonal architecture:
- Domain: entity types and LlmError enum
- Application: prompt building, sampler construction, tool call parsing
- Infrastructure: LlamaEngine wrapping llama-cpp-2 with isolated unsafe transmute
- Presentation: Axum handlers, middleware (auth), error chain, router
- Config: type-safe AppConfig with LazyLock
- Bootstrap: Application struct with build() + run()

Resolves build_sampler/build_sampler_params duplication.
Adds simple web chat UI at GET /.

Co-Authored-By: Claude Code <noreply@anthropic.com>

src/presentation/handler/health.rs

@@ -0,0 +1,13 @@
+//! Health check endpoint.
+
+use axum::Json;
+
+use crate::config::MODEL_ID;
+use crate::domain::entity::HealthResponse;
+
+pub async fn health_check() -> Json<HealthResponse> {
+    Json(HealthResponse {
+        status: "ok".into(),
+        model: format!("{MODEL_ID}-q4_k_m"),
+    })
+}