CHORE: Replaces socket.io with native websockets

[billchurch]

I believe socket.io has outlived its usefulness for this project and I think the bulk of the browsers used today support websockets natively.

I've created a new branch ws-experiment to test that change which I think we'll target as 0.5.0 for now.

In addition to this, I intend to drop support for node 6, and potentially even node 8 and make node 10 the minimum supported version going forward. It's become increasingly difficult to keep packages up to date and maintain support for those older versions of node.

[diogenes1oliveira]

Hello, @billchurch, I'm interested in testing or possibly developing the native websockets version, but I couldn't find the branch you've mentioned. If it's not too much to ask for, do you still have it somewhere?

[billchurch]

I did kill it as it was a little more involved than I was hoping at the time.

However, I would like to revisit it but need a little more planning around it as Socket.Io does abstract a lot of things for us.

Thoughts?

[diogenes1oliveira]

@billchurch sorry for the late response, specially compared to the promptness of yours...

Yeah, I was checking it out a little more and you'd need to come up with some encoding of messages to replace the socket.io events/channels. It wouldn't be too difficult, but maybe it's not worth it. Maybe just expose an option to set "transports:["websocket"] in the socket.io setting, excluding the initial long polling requests?

My motivation was mostly to transport some binary data as well through the UI, but I came up with a simple Go command to encode the streaming command input/output as base64.