Nick is correct on this one. The swash driver does not serve any purpose other than to set phasing and certainly doesn't play a role in determination of collective travel once it's verified that there is no binding. Having the arms at 90 degrees is not of critical importance as long as you keep in mind the two practical (mechanical) limits of the driver system below:
1. At full positive (negative for trailing edge control) collective on a leading edge-controlled rotor system and a corner collective setting (full-throw elevator + aileron), there should be no swashplate binding on the driver.
2. At full negative (positive for trailing edge control) collective on a leading edge-controlled rotor system and a corner collective setting (full-throw elevator + aileron), there should be no swashplate binding on the bearing block and your driver link arms should not be under load at extreme angles. In other words, the ball links should not be at the maximum extent of their travel, as they're pretty easy to pop off of the balls with a relatively small load and if they don't pop off, they will wear unevenly, which will result in eventual failure.
What's as important if not more than the above two items is to ensure that you phasing is zero'd out correctly so as to not induce any undesired control interactions. To clear up a pretty typical misunderstanding, FBL systems can NOT electronically correct for mechanically-incorrect phasing or artificially-induced phasing due to misaligned sensors... It's got to be perfect or you're going to feel it in flight.