How does electronic first curtain shutter match the second curtain?

I previously incorrectly
understood how the shutter curtains work. I thought they move at so high
speed that e.g. at 1/4000 s shutter speed, the first curtain first opens very
quickly, the shutter is open for 1/4000 s and the second curtain closes very
quickly.

However, I was recently told that at 1/4000 s shutter speed, the curtains move
really slow and there is only a small moving slit between the first and second
curtains.

My camera, a Canon EOS RP, however, has electronic first curtain shutter. The
second curtain is mechanical.

How does this electronic first curtain shutter actually work?

My first understanding of it was simply that it globally connects all of the
CMOS photosites to ground (so all pixels have the RST line active
using the terminology of
Wikipedia’s CMOS sensor page),
and then the RST is deactivated simultaneously across the whole
sensor area.

However, if the second curtain is moving slowly, obviously the first curtain
has to be moving slowly too.

So, my current understanding is that the RST lines are first
active, and then gradually deactivated on a row-by-row basis, so that the row
where reset is to be deactivated moves exactly at the same rate as the
mechanical second curtain moves.

Is this current understanding correct? I think this current understanding would
avoid the severe rolling shutter effect of electronic second curtain. The only
problem would be to match the movement of the electronic first curtain very
accurately to the movement of the mechanical second curtain.

This current understanding would also explain why the second curtain is
mechanical, i.e. CMOS photosites have no way to be stopped while maintaining
the current charge, apart from covering them with a mechanical curtain, and you
can’t read the whole sensor area in few milliseconds.