Retained, yes, but no linked physical judgement; the labels were added while looking at the pictures, which is exactly the bit I don't trust.

Get quality to establish the sample decisions first. I cannot tell whether those are defects from your description, and moving glare does not prove the underlying bezel is acceptable either.

Keep full capture settings and the seating arrangement with each image. A crop of the stripe won't explain the presentation change later.

Quality has the retained set now; I've withdrawn the scratch captions from the working references and kept the originals separately, with the capture notes we can recover.

When the decisions return, compare difficult accepted surfaces alongside identified rejects under the proposed presentation conditions. Removing the bright stripe alone would be a poor test if it also made the real rejected feature disappear.