Procedural face rendering

Cozmo .bin FlatBuffers animation files describe Cozmo face images procedurally. Each face image is described with a structure in the following format: https://github.com/zayfod/pycozmo/blob/1fbf2775c2147356b39a031be7d92d74bb223a16/pycozmo/CozmoAnim/cozmo_anim.fbs#L11

The left and the right eye are described with 19 coefficients each.

Code that can interpret procedural face descriptions is needed. It should take pycozmo.CozmoAnim.ProceduralFace object as input and should generate a 128x32 PIL image as output.

The following presentation from Anki provides some background information: https://www.gdcvault.com/play/1024488/Cozmo-Animation-Pipeline-for-a

[zayfod]

The latest commit on dev (5a62b146d2230741cda51153e4fc8c660ade65bd) finalizes the support for all 5+19+19=43 procedural face parameters.

There are now two examples:

• procedural_face.py - displays procedural faces on Cozmo’s face (128x32)
• procedural_face_show.py - displays procedural faces on the screen (128x64)

The next step is to integrate procedural face rendering with animation playback.

[zayfod]

With the coefficient interpretation from @ppedro74 , Pillow.ImageDraw, and the new Client.display_image() method, everything is in place to start working on procedural face rendering.

I am planning to add a new pycozmo.procedural_face module to dev shortly.