TOOL-519: MCP client POC

[the-nick-fischer]

This PR includes the following:

• Implement needed changes on the chipmunk core: TOOL-516
  • Chipmunk internal MCP API]: Thin MCP API that bundles creation and interaction with the MCP client and server:
    • create and manage MCP API instances
    • run the MCP event loop to process incoming messages
    • methods to send messages to the MCP client and server
    • => application/apps/indexer/session/src/mcp_api/mod.rs
  • Chipmunk session changes:
    • Instantiate and start MCP server and client.
    • Send a mock test prompt
    • => application/apps/indexer/session/src/session.rs
  • Chipmunk state API: Example changes showcasing where UI triggers would be connected to the MCP API => application/apps/indexer/session/src/state/mod.rs
• Implement a basic MCP client POC: TOOL-519
  • application/apps/indexer/session/src/state/mod.rs
  • new and start functions to create and spin up the MCP client (connecting to the MCP server)
  • run function that initializes conversation state and handles incoming messages (e.g. prompts form chipmunk) and processing of chat messages using the llm client
  • Abstraction layer for LLM clients => application/apps/indexer/mcp/src/client/llm/mod.rs
  • A mock LLM client that simulates LLM behavior without making actual API calls / without a HTTP client =>application/apps/indexer/mcp/src/client/llm/mock.rs

Note: Open topics or Notes are marked with // TODO:[MCP]
Note: contains log::warn!() statements for debug purposes. They will need to be removed / replaced by sensible debug statements.

Logical flow explained:

• Chipmunk is started. In session.new()
• The MCP server and client are instantiated
• The MCP server is started
• The MCP client is started
• A mock prompt message is sent from chipmunk core via MCP api mpsc channel to the MCP client
• The MCP client uses a mock LLM client to handle the prompt, which wants to make MCP tool call
• The MCP client sends a tool invocation to the MCP server
• The MCP server sends a tool / task request to the chipmunk core via mpsc channel
• Chipmunk handles the request by sending a Search request via operations API
• Result is returned to MCP server -> MCP client -> LLM client
• LLM client sends a Final response