trim video to exercise

[MasterMedo]

Create a python function with the following declaration:

trim_video_to_exercise(
    body_part_positions: list[list[list[float, float, float]]]
) -> tuple[int, int]

The function should take in a list of body part positions at all frames of the video and return the starting and ending frame at which the exercise is found.

The input list has as many elements as there are frames.
Elements of the initial list are body part position lists.
Body part position lists have 17 elements, one for each body part that is being detected.
Each element of the body part position list is [x coordinate, y coordinate, confidence].
Here is the example input:

[
    [
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
    ],
    # ... the rest of the frames
    [
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
        [0.4, 0.3, 0.8],
    ]
]

The output is a tuple of starting and ending frame.
Here is an example output:

(123, 1806)