There likely is no defined standard. Broadcast video signals have a minimum attenuation of -30 dB at the stopbands. But that's AM. Laserdisc is FM, so things are a bit different. The frequency deviation is not going to be any greater than 7.1 MHz. Broadcast video is limited to 4.2 MHz, with audio, 6 MHz. The carrier is quite low-frequency compared to the message signal. The transmission is also very low-noise. For this reason I suspect there is no hard band-limiting applied, and rather things are just left to their natural devices. This page seems to support this: http://machineryequipmentonline.com/vide...nal-processing/
The PWM is actually just a clipped sine and adds harmonics. These are so high-frequency compared to the carrier that they can be filtered out and do not interfere with the message at all. This is the reason FM is popular for music. The transmitted signal can be clipped with no loss of information, unlike AM. This is perfect for a disc format due to jitter between the laser head and the platter.
I think it should be sufficient to use a brickwall lowpass filter at 14.2 MHz, sample at least 28.4 MHz. There will likely be nothing but junk at the high frequencies, but it shouldn't be very noisy anyway.
From there you can either demodulate it into a standard NTSC or PAL composite video signal (or a video format) and store it, or store it as is.