I don't like very much, what bitscope designer says. He talks about 100MHz sub-sampling bandwidth, but uses an ADC showing a bandwidth of only 75MHz (typically). Then he says, that one can see components even higher than 100MHz. Of course, but you can do this with any other scope: You can always see components higher than the bandwidth, but highly damped of course. And if you have a look at the datasheets of chips and ADC used here, then you can see how sharp the signal falls beyond 75MHz and it's very unlikely that you can really see something useful at 100MHz or even beyond.
Bitscope works with some tricks, only typicals are given and high bandwidth is only achievable with certain gains, when the AD8048 is not in the signal chain. As a gain of about 5 is wanted from this chip it will relevantly decrease over all bandwidth and slew rate.
Another trick bitscope uses is to insert HF-decompensation networks. This is dangerous, because you must exactly know about the lack of bandwidth of chips to do this. But this parameter isn't stable at all, but depends on manufacturing tolerances, temperature, long term drift, etc. So, the one might be lucky and can achieve about 75MHz, the other might be unlucky and will result in some lower bandwidth or even a heavy peak somewhere. By fumbling with the compensation network you can see nearly everything, but you totally lose gain linearity.


Well Fourier theory tells, that this is possible.


I should see the whole circuit first.


Have a look into datasheet. Look at the gain versus frequency diagrams. You will notice, that the gain is given in dB here.

Kai