HLS is a way to serve a video in chunks over HTTP. It's how livestreams are implemented in most places. Twitter serves videos in HLS chunks, but Nitter doesn't want to proxy it to you because videos take a lot of bandwidth. In order to let you choose whether you want to connect to Twitter servers, it first asks you whether you want to enable playback.

Because they want to avoid sending your IP to Twitter if you aren't going to watch the video, and the way to do that without JS is to only render the video embed if you actually request it.