Whole application should stop if any message system throws an error

[ForNeVeR]

Currently this code manages the application lifetime:

emulsion/Emulsion/Program.fs

Lines 25 to 58 in 877144c

	let private startMessageSystem (system: IMessageSystem) receiver =
	Async.StartChild <| async {
	do! Async.SwitchToNewThread()
	try
	system.Run receiver
	with
	| ex -> logError ex
	}
	let private startApp config =
	async {
	printfn "Prepare system..."
	use system = ActorSystem.Create("emulsion")
	printfn "Prepare factories..."
	let restartContext = {
	cooldown = TimeSpan.FromSeconds(30.0) // TODO[F]: Customize through the config.
	logError = logError
	logMessage = logInfo
	}
	let! cancellationToken = Async.CancellationToken
	let xmpp = Xmpp.Client(restartContext, cancellationToken, config.xmpp)
	let telegram = Telegram.Client(restartContext, cancellationToken, config.telegram)
	let factories = { xmppFactory = Xmpp.spawn xmpp
	telegramFactory = Telegram.spawn telegram }
	printfn "Prepare Core..."
	let core = Core.spawn factories system "core"
	printfn "Starting message systems..."
	let! telegramSystem = startMessageSystem telegram core.Tell
	let! xmppSystem = startMessageSystem xmpp core.Tell
	printfn "Ready. Wait for termination..."
	do! Async.AwaitTask system.WhenTerminated
	printfn "Waiting for terminating of message systems..."
	do! telegramSystem
	do! xmppSystem

As you can see, nothing stops any message system from completely failing (i.e. kill the child async started in startMessageSystem), but the actor system (i.e. the whole app) will still live.

We need to stop the whole application on critical failures (e.g. dead message system).

Take into account that throwing an error from a system.Run is a very unusual event: usually the message system will just restart itself without terminating Run.