Overall I think you have missed the big picture. The idea of an approach is to land. Apart from a CAT IIIB landing, you have to have some sort of visual reference to land. The worse the visibility, the lower you have to go to see the runway. The lower you go, the greater the system integrity and training required. The very lowest minima flown to at the moment are CAT IIIB that require just 75 M RVR. You never ever have to be visual to perform a go around.
Q1 - CAT I Ops are performed to to a Decision Altitude, ie. based on barometric data and not radio altimeter. The minimal do not vary with LVP’s. These approaches can be flown by very basic aircraft. You can still autoland, but you have to make your decision based upon barometric minimums.
Q2 - With either a manual or autoland, you have to have the required visual references to continue beyond Decision Height. If conducting an autoland, any failure after DH is to likely to end up in a go around. That is because it is very difficult to “revert” lower than 100 above the runway. CAT II Manual landings may be required when the slope in the touchdown zone is beyond the aircraft’s limits.
Q3 - Improved airport infrastructure and system testing allows lower minima. Again, you have to have sufficient visual reference to continue beyond DH.
Q4 - The alert height is typically 100 feet above minimums no matter what the approach type, it has nothing to external or internal system integrity. There are also no guarantees that a systems will never fail, so you can always go around. But the reality is, weather limitations mean that if you have a system failure after selecting reverse you are at a point where the aircraft is rapidly losing speed and the aircraft is quite controllable. After all, you will still have 75 M RVR.