Implement multiprocessing for video inference and other cool stuff

[anth2o]

This MR implements the following features. I splitted commits so that each feature has its own commit, thus I advise you to review each commit at a time.

• multiprocessing for video inference (almost *5 in inference speed) #13
• thresholding per class after object detection and before tracking
• add box coordinates for each thrash in each frame in tracking output
• more documentation about serving, and add the possibility to send videos through curl
• improve readme and tests docker

[codecov[bot]]

Codecov Report

Merging #17 into master will increase coverage by 1.07%. The diff coverage is 100%.

@@            Coverage Diff            @@
##           master     #17      +/-   ##
=========================================
+ Coverage   94.23%   95.3%   +1.07%     
=========================================
  Files           5       5              
  Lines         208     213       +5     
=========================================
+ Hits          196     203       +7     
+ Misses         12      10       -2

Impacted Files	Coverage Δ
src/mot/tracker/video_utils.py	100% <ø> (+11.11%)	:arrow_up:
src/mot/serving/app.py	77.77% <100%> (ø)	:arrow_up:
src/mot/serving/inference.py	96.05% <100%> (ø)
src/mot/tracker/tracker.py	95.4% <100%> (+0.05%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 83702d6...f927af6. Read the comment docs.