We might be running on a different system from where the input was sent, or a notebook authored on one OS might be opened on another, so we should split on both \n and \r\n. I think we have code elsewhere doing that and it might be a worthwhile utility method to pull out.

Unrelated, what if the SubmitCode.Code contains newlines?

A single command/response must not contain newlines because we read them by calling console.ReadLine() (effectively). The code we're looking at here is the test code, which just pushes everything into a single stream and then reads it all back at once (hence the splitting)

What's the process when someone submits a cell with multiple lines of code inside it? If we read the input to the end, instead of line by line, could it simplify this case?

I think it's not valid json to have an unescaped line break in a string. It's fair for us to expect clients to escape their newlines.

Will add a test for this though