Restart Handling

[arlaneenalra]

Watches have a built in timeout that causes them to eventually expire and disconnect. This leads to conditions similar to #6. To work around this, we have a restart loop in the code and a Restart=always policy in the systemd unit right now. (Of course, that unit file is not presently part of this code base.) The real fix here is to define configurable restart handling in the service itself.

Here's a proposed approach:

• Add a configurable reconnect loop that supports max number of restarts, including unlimited.
• Add the systemd unit file to the code base to be packaged in the dist tarball.

It might make sense to setup deb/rpm packaging for this. (Packaging is covered in #9)