The only time it is somewhat justifiable to wait for a Firestarter proc is on the last Fire 1 of your fire cycle.
Here's the issue. You cast your last F1, then immediately queue B3. F1 lands, you get a FS proc, but B3 now has about 0.5s left on its cast, which is the time interval where you can move without breaking a cast... So your B3 goes off. Now you do your regen cycle, and are now at full mana with 2-3 sec left on your FS proc.
In this situation, your options are to use another B3 for 220 potency then F3 without the FS proc - perhaps leading into the double-F3 speed cast if you are good at it.
Or you can use F3 with the FS proc while in UI3 (154 potency), saving you 0 casting time (fast-cast F3 from UI3 vs Instant from FS), and a whopping 133 mana.
Now if you wait that extra half-second on the last F1 in your fire cycle, 40% of the time, you should get an extra F3 at 396 potency out of it.
A single cycle takes roughly 22 seconds, lets say you are very unlucky with procs and it takes 20 seconds. So by adding an extra 0.5s to your 20s cycle, your dps is dropping by 20/20.5 = ~2.5%.
The shorter cycle using the double-fast F3 trick has a net potency of roughly:
154(B3 in AF3) + 395(Full T2) + 154(F3 in UI3) + 396(F3 in AF3) + 270x3 (F1 in AF3) = roughly 1909 potency in ~ 20 seconds, with no procs.
This is ~95.5 potency/sec, so adding 0.5s to the cycle drops it to 93.1 pot/sec.
By adding this 0.5s delay, we now have a 40% chance to get an extra 396 pot F3 into the cycle, adding ~2.5s to it.
So 60% chance to have 1909/20.5 and 40% chance to have 2305/23 = 95.96 pot/sec average.
So ~95.5 with no procs and spamming buttons, or 96 with the 0.5s extra delay. In this case it IS slightly beneficial to wait for an FS proc on the very last F1... but not every F1. Granted the gain is small, but as we get more procs and have longer average rotations, the dps penalty of the 0.5s delay will get smaller as well.
I guess its a tossup... But uh, never wait on ALL F1s. Wait on the last F1, or none at all.