A few notes...
The conjunction screening volume for ISS isn't a cube. It's a rectangular box which is longer in-track and smallest in the radial direction.
It kind of sounds like you're suggesting this volume is so that the spacecraft has time to maneuver. This isn't the case - any maneuver to mitigate a predicted conjunction would have to occur long before the event - at least half a rev is good, more is better.
The reason you have a screening volume larger then the satellite is because the satellite orbits have a significant amount of uncertainty, commonly referred to as the covariance. So even if you predict a 2 km miss between two objects, there is still a non-zero chance of collision because of the position uncertainty. The iridium/cosmos collision had a predicted miss distance from TLE data of on the order of 0.5 to 1.5 km, depending on which data you use.
This is also the reason why conjunction predictions, even for ISS are only done a few days to a week out. The position covariance grows with time so eventually it's to big to do useful analysis.
So back to Mars. Space is big - really really big. 5 objects in orbit around one planet should have a vanishingly small chance of colliding. With over 17K tracked objects around Earth, collisions are still a very very rare occurrence - 4 events to date.
The "space is big" argument isn't very satisfying however, because these objects aren't placed there randomly and there are orbit altitudes and inclinations which are more useful then others, so you would expect that objects would be concentrated there.
Back to covariance - does anybody know the accuracy to which we know the orbital positions of these probes? Sounds like a good follow-up question. If the covariance of the objects becomes to large, you really can't do effective collision avoidance analysis.
The best strategy may be to simply allow a sufficient altitude difference between the missions - which shouldn't be difficult given the population now or in the foreseeable future. But the risk really is quite low.