in this case, it is important to be aware that the times you specify may not be extracted exactly. it will be off by a few frames based on keyframe availability. the only way to extract exact frames is to re-encode. :)
Personally I've created dozens of clips using this method and it always turns out ok. It gives you about ~1 second precision on where you want to make your cuts. After I create the clips I can play things back normally, complete with an ability to seek to specific points successfully.
Yes, in action recognition tasks (machine learning), e.g., if you have a large video with temporal annotations (start/end times where an action occurs) you may want to extract clips to sense-check the annotations. Being exact is important.
but the ~1 second precision that you see is by accident where the source file happens to have a keyframe every 1 second. that may not be the case always. :)
1. re-encode the whole thing
2. re-encode just the first GOP
3. start a bit early or late
4. include the full first GOP but use an edit list to instruct the player to skip to the timestamp of interest