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>

Cargo.toml

@@ -17,6 +17,9 @@ tower-http = { version = "0.6", features = ["cors", "trace"] }
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 
+# Error handling
+thiserror = "1"
+
 # Utilities
 tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }