Feature request: set staleness timeout value inside the token

[pelletencate]

We need ownership of stale timeout by the process that put the lock in place, i.e. have the lock itself be aware of its stale timeout value.

If process A requests a lock, and process B requests the same lock with a staleness timeout of 30 seconds, it will recycle the lock placed by A after 30 seconds.

It would be very useful if the timeout value could be stored inside the lock itself, so that process A could lock with a 30 seconds stale timeout, which process B can respect regardless of how it was called.

This is useful if multiple processes want to acquire the same lock for different reasons, with different timeouts. For example, in a web app, a web server process could be expected to lock for example longer than 2 seconds, whereas a background job may be expected to work on the same resource for significantly longer. In that case, you don't want the web server to be able to request short-lived locks without invalidating the longer-lived locks put in place by the background workers.