Add face landmark task

[vinlemon]

Fixes #7

This PR adds the Face Landmark task #7 .
The handling of output_face_blendshapes and output_facial_transformation_matrixes is still missing.

[hydai]

As I am not familiar with the face landmark task, I have a few questions, could you please help me to figure out?

[hydai]

Also, would you like to add a new example under README to show this fantastic feature?

[apepkuss]

@vinlemon Could you please provide an example in a markdown file, showing how to use the code to finish a face landmark task? The example can help users get familiar with your code and how to use them much more efficient. Thanks a lot!

[MediosZ]

The code LGTM, but I have a comment about the source of the model.
I notice there are two models for face_landmarker, one from mediapipe-models and the other from mediapipe-tasks. They have different number of outputs.
I think we should stick to one of them.
The script downloads the one from mediapipe-tasks but the one from mediapipe-models is more accessible, since it's on the official website

[vinlemon]

@hydai @apepkuss I've updated the README and added a separate file for the example, which should make the face landmark task clear and easy to follow. Let me know if anything else is needed.

[hydai]

LGTM. @apepkuss PTAL. If everything is good, then let's merge it :-)

[hydai]

Thanks @vinlemon