The previous visualizer.py file was too big and unreadable, so might lead to problems later when making changes if necessary. So, in this PR, I do the following:

1. Create a subfolder for the visualization and do everything there
2. Split the existing visualizer into its core components, such as the RegistrationVisualizer and ClosuresVisualizer and use them through the main Visualizer class, and wrap main API functions around this new pattern.
3. Create a common dataclass LocalMapData containing the individual local maps, density maps and other params relating to them, that can be shared across different visualizer components, without having to use up too much space.
4. Lastly, I added an important visualization feature to loop over all the local maps and corresponding density maps generated upto any given frame in the dataset. This is very useful for debugging the performance of MapClosures on challenging datasets. This is the LocalMapVisualizer.
5. Each of these individual classes are defined in separate python files for better maintainence.