Oh boy, this is going to be one of those mega threads. Firstly I'm glad you've brought this up and support the changes proposed so far, perhaps with some minor quibbles. I don't want too much added complexity but obviously the current system is a little bit too much handwavium.
1) Good idea, perhaps frozen water should also be taken into account when a planet reaches a suitable temperature? It would require determining how much surface various bodies in the solar system has that would actually be liberated once thawed. I think the moon has a fairly large surface water deposit, but certainly not enough to reach 20%, perhaps low gravity bodies with the minimum 2 colony cost should actually have their cost increased higher than 2 by factors which on other worlds only raises cost to 2? For instance low water availibility causing a cost of 3 or 4 on the moon rather than 2. Simple addition of 1 extra colony cost on top of all other factors for these bodies is probably easiest.
2) The mechanics of this should be interesting, At around 0.5%-1% side effects start to develop, causing symptoms of intoxication affecting memory, motor skills, critical thinking etc, above 5-7% for short periods can cause unconsciousness. Long term exposure to above 10% Co2 is generally fatal, though I suppose genetic modification might allow higher percentages. A reduction in wealth and industrial production at levels over 1% seems like a good idea to me, with population growth being affected when concentration gets high enough.
3) Makes sense, however there is another factor affecting small bodies mostly, and certain large bodies also. If a planet has never had an oxygen atmosphere then the surface will be covered in highly reactive compounds, like iron, which will oxidize as you add oxygen to the atmosphere. I read somewhere than you might need 3 times as much oxygen to be added to the moon before it will become breathable due to surface rust forming. Mars doesn't suffer this problem as It had significant oxygen as it's crust was forming causing surface iron to already be oxidized.
4) I can't find relevant numbers right now, but I've heard anywhere between 100-1000 years for a terraformed atmosphere to be stripped off the moon by solar wind. One interesting factor is that as gas escaped from a moon at least half of it will become part of a torus in orbit of the parent body, some will impact back into the satellite as it passes through, and a large proportion will fall back into the atmosphere of the parent.
Going by a depletion time of 100 years for the moon you might add a mechanic that 1% of gasses escape per year, so terraforming facilities would be needed to keep the atmosphere topped up. It would be frustrating to need to manually swap facilities around between adding nitrogen, oxygen, greenhouse gasses etc. So add a mechanic that allows existing facilitys to simply offset their production against lost gasses.
I'm in favor of slowing down terraforming, it really should be an insanely difficult process. We lucked out with mars being right next door though.