Bug description

It appears the implementation for StdioServerTransportProvider could potentially mismatch encodings between input and output.

The startOutboundProcessing method will always force the encoding to be UTF-8:

 synchronized (outputStream) {
	 outputStream.write(jsonMessage.getBytes(StandardCharsets.UTF_8));
	 outputStream.write("\n".getBytes(StandardCharsets.UTF_8));
	 outputStream.flush();
 }

However, the startInboundProcessing method doesn't appear to specify an encoding when creating the InputStreamReader object, and so it could be subject to the default charset selected by the JRE:

reader = new BufferedReader(new InputStreamReader(inputStream));

Environment
Testing with 0.10.0

Steps to reproduce

Running a STDIO-based MCP server with the JRE default charset set to something other than UTF-8.

Expected behavior
I'd expect charset/encoding handling should be consistent between input and output.