There are a lot of things that effect the time it takes to play a tournament to the end besides the time of each round. Starting blinds compared to starting stacks, adding antes at a certain point, the amount of the increase at each round, rebuys etc.
Surprisingly though, even if the number of players starting varies by a large amount, the time to play a tournament does not change by the same relative amount. The more players there are the higher the rate of players going broke is. Since each limit the *M's roughly double, doubling the number of players entering will generally/maybe add about one round to the time it takes to finish a tournament. It is almost like clock work. So I am thinking "Or is taking the amount of players into consideration a dumb idea to begin with?" while not dumb, not that big of an issue.
Stats I am aware of, just from being around a long time. In a full nine handed game of limit Texas hold-em, the hourly rate of hands averages about 35. Add a shuffle machine it goes to about 45 hands per hour. NLHE is a few hands an hour slower then limit. Playing online is faster, one site at least used to keep a running tally, I remember going over 120 hands playing head up. Stud poker was about 25 hands an hour. Other games like Omaha, draw games are generally slower then stud.
I like the ideal of being able to configure the tournament. I think preset choices are really a good ideal. There are generally three types of tournament structures. Turbo, regular and long play. Turbos are timed to generally get the button around the table once around, typically 10-15 minute rounds. Regular's are meant to get the button around 2 to 4 times, 20 to 40 minute rounds. Long play or deep-stack, longer then that, minimum 1 hour rounds. Do something like that and you will have most of what you need.
The biggest thing besides the length of each round, is the relationship of starting blinds to starting chips. (this is one of the things tournament directors tinker with the most, I have seen a few times where they double the starting blind and the amount of TC, for marketing purposes, or raise the starting blind to save payroll). What I would suggest is that you lookup some tournaments, look at the starting chips and blinds, and make some phone calls. They might be vague about starting chips and blinds with their advertising, but call and ask and they will be happy to tell you. Than ask how long the tournament typically lasts. In Las Vegas, Call Sam's town, they run three tournaments a day, Orleans runs one or two. Look at online tournaments for the same type of thing. You won't come up with much specific data but you will get some general benchmark kind of stuff.
Hand per hand play, has its downside or more accurately may not be quite ready for primetime. Outside of larger events, mostly at the bubble I have not seen it done. I am not sure it would be more elegant then time, nor am I sure that it would be more fair then the time clock. People think a hand count of say each player sees say ten hands a level and base that on ten players at the table. That's assuming there are always going be ten players at the table which there rarely is outside of the first hand of the tournament. So there seems to be about the same number of inequities with hand per hand as there is with time. Player are still going to get the advantage of the button twice, or the blind twice, and tables will still need to be kept balanced to a standard, just like with timed rounds. About the only advantage I see is screwing up the duffess trying to slowdown the game so they keep some position after the blinds increase. I am shooting from the hip on this, would love to hear about how your thinking about doing this.
Besides the presets I would suggest a snippet about how your users can modify certain things to make their tournaments play longer or shorter.
*M,s are the cost of each round to the players in blinds and antes, from Dan Harrington's books.