I found the video on Youtube. The Rigol test starts at 16:26. If I saw that transient response, I would consider the oscilloscope broken or in need of calibration.
I remembered wrong though. It is an actual MSO1074Z and not a hacked 1054 but that just makes the situation worse.
I don't have a MS 1074Z, or a source of fast rise time signals, so can't reproduce the result, but the rise time looks reasonable to me, just "counting squares" on the pix, & the bit of pre ringing & overshoot aren't all that dire.
The problem is not the rise time although various users have reported a wide range of rise times at different signal levels which is consistent with the problem shown. The rise time should be invariant with signal level.
The preshoot is normal for a DSO where sin(x)/x reconstruction is being used and aliasing is present. You just have to live with it if the sample rate is insufficient for the bandwidth. If this test was done with more channels active lowering the sample rate to 500 MS/s or 250 MS/s, then the preshoot would be worse.
The problem is that straight slope after the rising edge. It indicates that the vertical amplifier chain is recovering from a signal which exceeded its full power bandwidth which should not be possible.
It would be instructive to test using rising and falling edges to see if that distortion is symmetrical.
I wonder if the 70 and 50 MHz models display the same behavior.