Is it something to do with air resistance?
There's a small effect for these smallish velocities, but that's probably not what you're seeing. The terminal velocity of a tennis ball is about 30 m/s. At the bottom of your 200 cm drop, the ball is going a bit over 1/5 of terminal velocity. Since drag is proportional to the velocity squared, it's a small effect. If you were dropping the ball from 100 meters vs a kilometer, drag would be a big effect. The ball would bounce about the same height for those large drop heights.
Theoretically, shouldn't the ball have the same bounce efficiency no matter the height?
In a spherical cow theory (the canonical example of an overly simplified physics model), yes. In practice, no. The coefficient of restitution is velocity dependent. Drag is but one factor for why this is the case. The ball undergoes some rather significant and nonlinear distortions during the bounce.
You measured a bounce of 102 cm for a 200 cm drop. That means your ball isn't legal. A legal tennis ball has to have a 138 to 151 cm rebound from a hard surface when dropped from a height of 254 cm under carefully controlled lab conditions. Your ball is a bit too lossy, so it's possible that what you measured is correct (for that ball).
There are many ways in which a physics experiment can go awry. Did you make multiple measurements? Did you use the same conditions (e.g., same ball, same temperature of the ball and of the air, same relative humidity) for each drop test? Does your release mechanism introduce a bias? Does your rebound measurement introduce a bias? Addressing the issues that can cause an experiment to go awry is very important with physics experiments.