Feature: add FaceRepository that does facial recognition

[YifanZhang05]

Key features

FaceRepositoy.add_faces(self, namespace: str, chunk_id: str, faces: list[Face]):
add the given face embeddings to Pinecone under namespace, with chunk_id (which video chunk the embedded face appears in) and face_id (which face it is) in metadata.
returns a dictionary where key is face_id, value is number of times appearing

add_images(self, namespace: str, chunk_id: str, img_lst: list):
first detect all faces from each image in img_lst, then call add_faces
returns a dictionary where key is face_id, value is number of times appearing

[YifanZhang05]

added chunk faces connector that stores face information for each chunk in modal dict

The dict structure looks like this:
{
chunk_id1: {face_id_1: 1, face_id_2: 1},
chunk_id2: {face_id_1: 2},
...
}

[eshaan-mehta]

LGTM in terms of functionality, but should make another ticket for testing

[eshaan-mehta]

should add more unit tests for each of the functions. It's ok for now, but you should add this as a next step as another PR