The sample you got there has become quite brown'ish. I'm guessing the transmissivity (is that a word?) was better when it was new.
That was my thought too. The sample is taken from the broken bits and I might suspect that it's broken because the layers have come apart. The (relatively) clear good sections aren't broken and aren't pickable. If the layers are coming adrift then all sorts of much will get in there and get in the way of light.
I'm actually very surprised that Dave seemed surprised that the sample was clear where the 'rough stuff' had come off - the video of that section of the roadway showed the non-borked parts were reasonably transparent and the brown bits that could be taken away almost opaque.
I think this kind of trashing is below the belt. Imagine if EEVBlog got reviewed and the consensus was that it was pants because you can hardly navigate the site and it's very slow to load, simply because the reviewer had chosen those times when Cloudflare is throwing a wobbly as the sample times. That's effectively what Dave did here, in my view.