Oh yeah, I'm sure the casino dice test is rigorous and professional, and probably linked to a manufacturing QA procedure.
"a cube of 36 chessex dice" is really a single anecdotal data point, though. 30 cubes of chessex dice would be better, especially if they were sourced to different runs.
That being said, the variance from random is so great that we can infer a manufacturing process that lets very poor dice slip through.
Too bad casinos don't use other polyhedra. For now, I'll stick with online rollers, and stay way from games that are overly reliant on d6s.