I have a manual positioning system for my solar panels, and I would say that difference between not positioned and correctly positioned panel is 2-3 times - in the morning and afternoons especially in winter time.
Yes, when the angle of incidence is bad that's true. But you have to average that over time to make a good comparison. Basically the power output of the panels drops off as the co-sine of the angle of incidence. So over a whole day you typically lose between 10 and 20%, take into account that in early morning and late afternoon 'twice the output' is not the same as twice the output around noon, and of course you are positioning your panels in such a way that they perform at their best when the lightpath through the atmosphere is the shortest (12:00).
So a 57% increase in daily power, in theory, near the equator.
But - I have ignored atmospheric absorption - at sunrise, the light goes through more atmosphere to get to your panels - I don't know how much this matters. On slightly overcast days, this is presumably very important - how cloudy is it where you are, in theory? :)
Also, you get power from scattered sunlight. Very roughly it is 10% of your panel's rated power, even when it has no direct sunlight falling on it. These 2 effects reduce the benefit of tracking vs static.
When I was looking at making my own tracking system, I found a Canadian company that claimed (iirc) "30% gain" for their tracking system. I thought 'only 30%, it seems hardly worth it?' (Except for load-balancing reasons).
Other stuff: Most domestic panels are on roofs - they are often not 'ideally sited' - e.g. do not face due south, roofs not steep enough.
That's true for a single panel, but you are overstates the gain for a panel farm because it ignores shadows cast from one panel into the others, that are extremely relevant early in the morning and late in the evening (exactly the times that tracking gains are biggest).
~95% iirc assuming 'perfect' cells. There are some very impressive triple layer panels out there that do 44%+ right now, these are typically used in space applications.
The balance between incident radiation and power produced is an interesting one, a 'perfect' solar panel would remain cold. Solar panels tend to produce less power as they warm up, so the better a solar panel gets at converting power the less it will get warmer! So there is a positive feedback mechanism at work here.
Typical run-of-the-mill cells are 25% (single junction), and on cold clear winter days they can keep that up for quite a while. So that's 250 Watts per square meter of panel.
I did some simulations to figure out the exact gain you could get from perfect seasonal and day tracking, it's about 28%, but no tracking system will ever reach that in practice. Typical reasons why you won't get to perfect: tracking systems tend to hunt the clouds if they are not clock based, otoh if they are clock based they tend to perform badly because they use energy to move when they're not producing energy themselves!