fix: make Context logging methods spec-compliant by accepting Any data type

[Jimil-Joshi]

Summary

Fixes #397

Per the MCP specification, the data field in LoggingMessageNotificationParams allows any JSON serializable type:

The data to be logged, such as a string message or an object. Any JSON serializable type is allowed here.

However, the Context.log() and convenience methods (debug, info, warning, error) were typed as message: str, which prevented users from logging structured data like dicts, lists, numbers, etc.

Changes

• Context.log(): Changed message: str parameter to data: Any to match the MCP spec and the underlying ServerSession.send_log_message(data: Any)
• Convenience methods (debug, info, warning, error): Updated to accept data: Any instead of message: str
• Removed extra parameter: Structured data can now be passed directly as data (e.g., ctx.info({"event": "processing", "count": 5})) making the extra workaround unnecessary
• Updated docstrings: Document that any JSON serializable type is accepted
• Added test: Verifies logging of dict, list, number, boolean, and None types

Example

`python
@server.tool()
async def my_tool(x: int, ctx: Context) -> str:
# String messages still work (backward compatible)
await ctx.info(f"Processing {x}")

# Now also supports structured data per MCP spec
await ctx.info({"event": "processing", "input": x})
await ctx.debug(["step1", "step2", "step3"])
await ctx.warning(42)
return "done"

`

Test Results

All 87 tests in tests/server/mcpserver/test_server.py pass, including the new structured data logging test.