Improve YOLO parsing errors

[zhiltsov-max]

Summary

Depends on #686

• Added more informative YOLO parsing errors
• Added more type annotations in extractor
• Allowed leading spaces in the .data file
• Added type docs for util.image.ImageMeta

How to test

Checklist

• I submit my changes into the develop branch
• I have added description of my changes into CHANGELOG
• I have updated the documentation accordingly
• I have added tests to cover my changes
• I have linked related issues

License

• I submit my code changes under the same MIT License that covers the project.
  Feel free to contact the maintainers if that's a concern.
• I have updated the license header for each file (see an example below)

# Copyright (C) 2021 Intel Corporation
#
# SPDX-License-Identifier: MIT

[IRDonch]

This enables questionable semantics: if any items are skipped in this way, then the number of items yielded will be different from the value returned by len(subset).

[zhiltsov-max]

I agree about this, but it is not clear how to resolve the problem, and it has very limited ways to impact the situation, because it is only used until the imported dataset is cached. During caching/iteration, the actual count is computed. Actually, in VOC we have the same issue. Maybe, we could even consider it normal, because unless we loaded the data, it is the best we can report.

[IRDonch]

IMO, the proper solution would involve removing __len__ from IExtractor.