Feat dataset generation

[SamFlt]

This PR introduces a new script, that uses blenderproc to generate synthetic data from 3D models. This tool can be used to train a YoloV7 to detect certain objects. Other modalities, such as poses, depth, normal maps and segmentation can also be recovered.

The accompanying tutorial explains the different tunable parameters, how to run the script and how to use its output to train a YoloV7-tiny to detect the objects.

Before merging, some result videos should be uploaded to Youtube and added to the documentation to showcase the detection results.

[fspindle]

Ready to merge after final check

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: -0.53% ⚠️

Comparison is base (2e4f1a9) 54.85% compared to head (0ced02b) 54.32%.
Report is 117 commits behind head on master.

❗ Current head 0ced02b differs from pull request most recent head 949c940. Consider uploading reports for the commit 949c940 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1217      +/-   ##
==========================================
- Coverage   54.85%   54.32%   -0.53%     
==========================================
  Files         399      405       +6     
  Lines       49139    49821     +682     
==========================================
+ Hits        26955    27066     +111     
- Misses      22184    22755     +571     

see 30 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.