docs: improve clarity around mcp examples

[jenna-winkler]

Is there a need for all 3 of these examples? if so, can we make the purpose behind each more clear in the titles?

• examples/tools/mcp_slack_agent.py (newest and used in tutorial)
• examples/tools/mcp_agent.py
• examples/tools/mcp_tool_creation.py

[planetf1]

It seems like mcp_agent is the nicest for understanding what's going on, since it has good observability, and runs a chat loop.
I can see why there's a simplified version in the tutorial - adding the extra code to handle events can detract from the message, as does looping through.
The examples could potentially be refactored but this might add confusion?
Not sure that tool creation in itself adds anything extra?

[planetf1]

Having a tutorial/example on accessing MCP server tools remotely (ie sse), as with the current proposed RemoteAgent seems like a good next step, as it further demonstrates the use of MCP as providing separation between the agent, llm, tools .