Move the normalization of the timestamps in preprocessing

[tizianoGuadagnino]

Motivation

The user or a dataloader should not handle timestamp normalization; this should be part of our pipeline. My opinion is that our system should process the data as raw as possible without requiring external manipulations. Furthermore, with this change, we can remove potential sources of errors as the user might not necessarily be aware that timestamps need to be in the range $[0, 1]$.

This PR

Introduction of a timestamp normalization step within the Preprocessor. This computation is performed just if deskew=True and timestamps are not empty. As a result, timestamps normalization is not required anymore in the generic and ROS-based dataloaders (mcap and rosbag through the tools/point_cloud2.py file).