It's definitely possible in my book.

Thanks for the link to that page of QSL cards. It looks like most if not all of those contacts are CW (Morse code). It is possible to detect bounced signals like that that are only a few hertz wide, but that may not scale up to the Apollo FM video, which was up to 2 MHz wide, or even more factoring in the sidebands. I have come across accounts of voice and teletype--only a few kilohertz wide, but nothing wider. The signal-to-noise ratio is much lower with the wider bandwidth video.

Video demodulators need a good signal in order to lock on to the video signal. The human ear is much more forgiving in picking out beeps buried in static.

Also, as I understand it, the varying roughness of the surface and the fact that the earth and moon are constantly changing their orientations relative to each other means that EME communications tends to be fleeting, with the signals fading in and out.