Hrmmm so it does... why do all the online resources have the wrong duration timers then? gah! Well that messes everything up...
I think I found a more accurate resource will rewrite my other post...
Edit again: I actually managed to find a nice breakdown of MP costs which should help in determining how much mana you eat up thus how much fire you can actually cast
Based on this, when you hit Astral Fire 3 (which you should always be at before casting Fire) it will cost you 638 mana. This means based on just my own personal mana pool (and I do not have anywhere near the best gear... but it is getting there) I can cast at the most 5 Fire spells 3670 total Mana - 638 * 5 = 480 -> blizzard 3 @ 79 MP = 401 -> Thunder 2 @ 319 (if you are using thunder 1 it costs even less mana so I have more than enough overhead).
So, lets try this again... Thunder @ 18s
Fire 3 (1.75) - Fire (1.5 - I still don't know how I get a short cast fire here... but I don't complain) - Fire (2.5) - Fire (2.5) - Fire (2.5) - Fire (2.5) - Blizzard 3 (1.75) - Thunder (2.5) = 17.5
So IF you get absolutely 0 procs you would be refreshing thunder a little early, clipping one dot tick. At 1 Proc you would have your DoT dropped off for 1 second, at 2 Procs the DoT would have been off for 2.5 seconds. At that point you have reached argumentatively a point where it is no longer profitable to hit Thunder 1. If you have a larger mana pool (Keeping in mind I am not full darklight yet, nor do I have my relic) and you get 1 extra Fire spell in there, then you are already hitting that breaking point of Thunder 1 with 0 procs.
Keeping in mind a chance of 40% to proc firestarter, after 5 fires your chances of not getting at least 1 proc is very small. I am just not seeing the benefit here, when all you are "sacrificing" for thunder 2 is .5 seconds... You get a higher initial hit (an increase of 20 potency), and you get just enough duration to make up for it. I think this is where I can buy the argument against using Thunder 3 at the moment, since we just don't have a large enough mana pool... yet. When we get enough mana, it will likely be better to use Thunder 3.
Edit 2: And if we start trying to sort out working in a Flare (I might try to napkin math this out to sort out if you should do this single target or not... although double flare would be worth it, I am not totally sold on single flare for single target) you are looking at a 4 second cast time. So at the end of my chain, you would be doing 5 fires, then a Flare, then a transpose (this is feels like around 1 second loss, anyone got a better number here?), then a Blizzard 3 (or maybe blizzard 1?) at full cast time (3.5), and then a thunder (2.5)... you have lost Thunder a LONG time ago (0 procs you are looking at a minimum of 23.75 seconds by this point you are risking a loss even on Thunder 2)