Hope you had a nice trip.

Flanking at Cannae:
Hmm - but at Cannae, the Romans lost 20,000-50,000 to 4,000-8,000. How can such one-sided killing be possible? The only way it would seem to make sense, is if the Romans had a complete morale collapse after becoming surrounded. Sabin has a very nice graphic description of what it might have been like in one of his pages.
So I think the evidence remains strong that outflanking was one of the best ways to get an army to give up the fight. It is so in Imperium, in that outflanked units have significantly worsened morale. Achieving an outflank, however, is not quite so simple. Cavalry are good at it. Infantry are pretty bad at it.
Wheeling is a very thorny issue - primarily, because it is extremely hard to model it in the game. I actually haven't reached a final decision on this yet.
Essentially, the basic moves of infantry (and cavalry) are to march straight ahead, march in inclination (i.e., forward and to the right or left), and the "about-turn" (probably will be done graphically with some sort of a counter-march). You will thus see very quickly that it is relatively easy for a unit to pursue its opponent and (if able to stop the pursuit), about turn to take enemy units in the rear.
Wheeling, on the other hand, is a thorny issue. Imagine the problems of turning a 2000 man long line. The current solution I'm playing with, works on the following premise.
From line formation:
1. The units form a "column" formation.
2. The units face about to the right or left.
This is obviously a rather slow and laborous process - stretching over 2, possibly more turns for infantry. Cavalry will be much faster at it than infantry, thus allowing them to be able to outflank relatively effectively. Drilled infantry will also be much more effective at this than non-drilled infantry. One way of working an outflank would be (like Hannibal) to deploy one's troops on the flanks in column formation, thereby ensuring that it would be a simple matter (1 turn) of facing to the right or left. Whether I'll be able to get the AI to use such a tactic is another issue entirely - for that matter, I'm not sure I should get it to do so.
One of the main issues I'm considering right now, is e.g., whether to permit "barbarian" units to make quick wheels in some simpler manner, at the cost of disorganisation. But then again, where is the evidence for that sort of thing happening?
But this will probably be one of those things that the open test will look to decide.