Handling of timestamps for different nodes/subject parts

[h-mayorquin]

This is a question that I think can help to document the schema better.

The schema says the following:

ndx-pose/spec/ndx-pose.extensions.yaml

Lines 72 to 74 in 8017291

	doc: Group that holds estimated position data for multiple body parts, computed
	from the same video with the same tool/algorithm. The timestamps of each child
	PoseEstimationSeries type should be the same.

But when I was working on sleap-io I remember finding on the example data that the algorithm did not produce estimations for some frames. What I ended up writing (because I did not read that part of the schema and the API did not throw an error) is different timestamps per node see here:

https://github.com/talmolab/sleap-io/blob/2bc3d5210c46bdb25413d25970c4bdc7adb6e8cc/sleap_io/io/nwb.py#L439-L441

As I was reading the schema today for totally unrelated reasons this jumped to mind and I just wanted to document it.

Is there a reason this is the expected behavior (same timestamps)? I guess it makes some analysis easier and it avoids some data duplication at the cost of forcing the users of the API to do some more leg work than before.

If this is the case should we?

1. Emphasize this point in the document?
2. Throw an error in the API if they are not? (is that possible?)

I can volunteer to add a small section in the documentation about this as a good practice.

Am I missing something?