There is more to phase change than just what you see on that graph. At temperatures and pressures that support multiple states (e.g. solid and gas) you have to look at the saturation vapor pressure of the gas and the actual vapor pressure (partial pressure) of the gas -- the line on the graph represents equilibrium between the states. The differences between the two will tell you if you have net sublimation or net deposition and determine whether you end up with a solid or a gas. For solid CO2 to form you need to have more deposition (transition from gas to solid) than sublimation. The transition to solid may also need help nucleating, though I'm not familiar enough with CO2 to say for sure.
To summarize, in your example Vostok at -90 C and 1000 hPa you have a rate of CO2 molecules depositing and a rate of CO2 molecules sublimating. The partial pressure of CO2 in the atmosphere is low enough that there is more sublimation than deposition, e.g. net sublimation. To deposit CO2 ice on the ground and have it stay there you would need atmospheric CO2 concentrations at least at saturation, which we do not have.
@Tristan provided some numbers in the comments that atmospheric CO2 vapor pressures (partial pressure) are <1 mm Hg (or less than about 133 Pa). To get net sublimation you need to either reach saturation via increasing the vapor pressure (add more CO2) or reducing saturation vapor pressure (lower the temperature). He provided an estimate that given CO2 partial pressures temperatures approaching -140 C would be needed for net deposition and CO2 snow.