Branch p8 (Avoid duplicate updates).

Waiting on active PR before submitting this PR.

The current implementation schedules an update at a fixed time interval (scan_interval), this leads to a couple of problems:

In case of reconnecting the updates stack up, and are executed fast (without the configured delay) when the connection is established, which will cause problem in many devices, and most surely will cause a spike in the statistics.

If a device is slow to respond, the scan_interval is not waited, which can lead to bad responses from the device.

This PR secures that each update is done at MINIMUM (might be more due to reconnect) scan_interval.