worker: claims fail to restart after connection to lightning is lost #993

Not sure if this is real yet, but it looks like:

• Start lightning
• Start the worker - it'll happily start claiming
• Start a run
• Kill lightning for a moment
• Restart lightning
• The run should complete, but the worker won't resume claiming

I think this might only occur if we're at capacity

This might just be a problem in my mock