It was already said above, but just to confirm, we try to detect black bars during video import, not during playback. This means only one value will be stored, which is the "biggest" active image rectangle found at the spots we are checking (which is random spots within the first 30 minutes of the video, I believe).
This is mostly only relevant if you have a ultrawide screen setup, either with eg. a 21:9 display, or a projector with an anamorphic lens to achieve it.
JRVR will use this data in a future update to automatically take the black bars off, and allow the image to fill such a ultrawide screen. For now we're only gathering data for future use.
"Live" black bar detection might come in the future, but as mattkhan said above, changing the rendering based on a live detection change might be rather jarring and unwanted.