I know it sounds simple But simple things are usually the most difficult things lol.
I wish there was more to suggest for this. But the way that the collections operate...I honestly can't think of a creative way to do this with what is currently built. There's no way for a collection to know how to reach out and play the video after it is tapped open...I think all it sees is a floating box. Kind of like you can't open a book and read it when it's sitting inside of a drawer that may or may not contain a book at all
I'm just a user here, I'm not on staff, so I can't answer to any further functionality that the IF staff could "Build" to do what you want.
But I can tell you what would work "if" the following feature was created:
If they programmed a way to cause a "Tap" at a pre-defined, exact location....which is something I'm actually going to suggest in a new post. If that were built into IF, you could program a timed-tap at right on the video, a split second after it opens.
But I might as well ask the obvious question...would it be Ok for the user to either tap on the play button once the video is open? Or to play the video directly from the gallery without opening it to float?