Spawn server

For complicated reasons, spawning subprocesses in Electron applications can block the event loop and impact application responsiveness.

The goal of this repository is to create a lightweight, high-performance server that Electron apps can use to spawn subprocesses on their behalf.

The server will listen on a domain socket.
When the client needs to spawn a subprocess, it will connect to the server via the domain socket and send the server JSON with a command path, an args array, and an env object.
The server will spawn the subprocess and stream chunks of stdout and stderr back to the client as they become available. When the process terminates, it will send back the exit code.

The goal is for the server to be a native executable, have a minimal memory footprint, and be completely event driven so it only consumes a single thread.

Currently, I have a toy example started in Rust that uses the tokio framework for asynchronous I/O. It currently listens on a TCP socket and just runs sh -c echo hello for every connection, streaming stdout back to the client.

Next steps:

Replace the hard-coded command with a JSON request supplied by the connection.
Stream both stdout and stderr back to the client with some sort of framing scheme that indicates which stream a given chunk of data belongs to.
Send the result code on process termination before closing the connection.
Listen on domain sockets on Unix, named pipes on Windows.
Write a client implementation in JS.