With a base probability of success of P, advantage is 1-(1-P)^2 or 2P-P^2. Disadvantage is P^2.
K advantage and (1-K) disadvantage is then K(2P-P^2) + (1-K)P^2 = 2KP - KP^2 + P^2 - KP^2 = 2KP + (1-2K)P^2.
At K=0.5 this is P, aka no advantage or disadvantage.
At K=1 this is 2P-P^2, aka advantage.
At K=0 this is P^2, aka disadvantage.
d/dK is 2P - 2P^2 = 2P(1-P). This is always positive, and maximal at P=0.5 (where it is 1/2) and zero at P=1 and 0.
What this means is that for any case besides certain failure/success, Advantage > Edge > Normal > Handicap > Disadvantage in terms of probability of success.
By having double-odd be bad on Edge, it doesn't block crits; your crit chance is doubled. Your natural 1 chance is halved over a normal roll.
With advantage, your natural 1 chance is basically eliminated (1 in 400).
Similarly, for Handicap, your natural 1 chance is doubled (just like disadvantage), and your crit chance is halved over a normal roll.
With disadvantage, your crit chance is basically eliminated (1 in 400).
This assumes crit on a 20; on a 19-20 the math changes a bit.
Chance of a 20: 1- .95 * .95 = 9.75%.
Chance of a 19 but not a 20: Die 1 lands on an even value that isn't 20 (9/20 chance) times die 2 lands on 19 (1/20), and vice versa (*2), for 18/400, plus double 19 (1/400).
Total: 14.5%
With real advantage it would be 19%.
With a normal roll it would be 10%.
Crit on a 18-20 with edge.
Same for 20 and 19.
For an 18 but not 19 or 20, it is die 1 on an 18, plus die 2 from 1 to 18, plus die 2 on 18 and die 1 from 1 to 17, for 35/400.
Total is 23.25%.
With real advantage it is 27.75%
With a normal roll it is 15%
again, about half way in between.