ai: add llamacpp backend (llama-swap) as third LLM client

Wires a new LlamaCppClient (OpenAI-compatible /v1 wire format) alongside
OllamaClient and OpenRouterClient. Per-slot routing for chat/vision/embed
via env (LLAMA_SWAP_URL + *_MODEL vars); capability inference uses an
env allowlist since /v1/models doesn't report modality.

InsightGenerator + InsightChatService gain three-way dispatch on
chat_backend = "local" | "hybrid" | "llamacpp". Hybrid and llamacpp
share the describe-then-inline path (text-only chat after a separate
vision describe). HYBRID_VISION_BACKEND=llamacpp lets hybrid route its
describe pass through llama-swap's vision slot while chat still goes
to OpenRouter.

Cross-replay matrix added (validate_cross_replay): local<->llamacpp
and hybrid<->llamacpp allowed; local->hybrid and llamacpp->hybrid
rejected. New /insights/llamacpp/models handler mirrors the OpenRouter
shape.

src/ai/mod.rs

@@ -5,6 +5,7 @@ pub mod face_client;
 pub mod handlers;
 pub mod insight_chat;
 pub mod insight_generator;
+pub mod llamacpp;
 pub mod llm_client;
 pub mod ollama;
 pub mod openrouter;
@@ -20,7 +21,8 @@ pub use handlers::{
     chat_history_handler, chat_rewind_handler, chat_stream_handler, chat_turn_handler,
     delete_insight_handler, export_training_data_handler, generate_agentic_insight_handler,
     generate_insight_handler, get_all_insights_handler, get_available_models_handler,
-    get_insight_handler, get_openrouter_models_handler, rate_insight_handler,
+    get_insight_handler, get_llamacpp_models_handler, get_openrouter_models_handler,
+    rate_insight_handler,
 };
 pub use insight_generator::InsightGenerator;
 #[allow(unused_imports)]