Refactor project structure and enhance backend integration

- Expanded package inclusion in `pyproject.toml` to support new modules.
- Introduced new `adapters` and `protocol` packages for better organization.
- Added backend adapter implementations for control plane integration.
- Updated main application imports to reflect new package structure.
- Removed deprecated core components and adjusted documentation accordingly.
- Enhanced architecture documentation to clarify the new runtime and integration layers.

engine/runtime/ports/llm.py

@@ -0,0 +1,67 @@
+"""LLM extension port contracts."""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Any, AsyncIterator, Awaitable, Callable, Dict, List, Optional, Protocol
+
+from providers.common.base import LLMMessage, LLMStreamEvent
+
+KnowledgeRetrieverFn = Callable[..., Awaitable[List[Dict[str, Any]]]]
+
+
+@dataclass(frozen=True)
+class LLMServiceSpec:
+    """Resolved runtime configuration for LLM service creation."""
+
+    provider: str
+    model: str
+    api_key: Optional[str] = None
+    base_url: Optional[str] = None
+    system_prompt: Optional[str] = None
+    temperature: float = 0.7
+    knowledge_config: Dict[str, Any] = field(default_factory=dict)
+    knowledge_searcher: Optional[KnowledgeRetrieverFn] = None
+
+
+class LLMPort(Protocol):
+    """Port for LLM providers."""
+
+    async def connect(self) -> None:
+        """Establish connection to LLM provider."""
+
+    async def disconnect(self) -> None:
+        """Release LLM resources."""
+
+    async def generate(
+        self,
+        messages: List[LLMMessage],
+        temperature: float = 0.7,
+        max_tokens: Optional[int] = None,
+    ) -> str:
+        """Generate a complete assistant response."""
+
+    async def generate_stream(
+        self,
+        messages: List[LLMMessage],
+        temperature: float = 0.7,
+        max_tokens: Optional[int] = None,
+    ) -> AsyncIterator[LLMStreamEvent]:
+        """Generate streaming assistant response events."""
+
+
+class LLMCancellable(Protocol):
+    """Optional extension for interrupting in-flight LLM generation."""
+
+    def cancel(self) -> None:
+        """Cancel an in-flight generation request."""
+
+
+class LLMRuntimeConfigurable(Protocol):
+    """Optional extension for runtime config updates."""
+
+    def set_knowledge_config(self, config: Optional[Dict[str, Any]]) -> None:
+        """Apply runtime knowledge retrieval settings."""
+
+    def set_tool_schemas(self, schemas: Optional[List[Dict[str, Any]]]) -> None:
+        """Apply runtime tool schemas used for tool calling."""