make Lock operations retriable

[phanky1]

Right now, the retry_period config doesn't apply to the self.lock object so if the Redis connection breaks right before tick() runs, self.lock.extend() throws an exception and beat stops.

This PR makes lock operations retriable when retry_period is set.

[sibson]

Is there a reason the getattr() isn't properly wrapping lock() to be retry-able?

[phanky1]

Is there a reason the getattr() isn't properly wrapping lock() to be retry-able?

it's because getattr() only wraps first-level method calls on the RetryingConnection. The returned objects of these method calls aren't bound to it