implement replay mode + factorize kafka seek offset from timing information

We had a technical break in our local cluster in Japan. The low-latency replication stopped for few days. This update aims to retrieve the missing files automatically within retention time range.

I have implemented a replay mode, seeking for the closest timestamp based on the requested replay time. A second option is used to limit the duration of the replay based on the filename GPS time.

NB: I am not sure why pipeline is failing. Looking at the logs it looks like some package missing ?