Bob,
Basically you are right that simply reshuffling the pack from a 3s to a 4s--without changing the motor too--will give you problems. However lowering the motor kV to adapt to the 4s pack won't bring you any gains either. I assume you want more speed at the high end, and maybe a little more capacity --but no more weight either.
So to illustrate the problem: Compare a 3s2600 pack (20C) to a 4s1950 (20C) pack ( both carry the same energy (3x2600=4x1950), the 4s pack max current is ~39A while the 2600 max current is 52A (close to your 55A static draw). Both can supply the same output watts (volts*amps), and if the individual cells were made with the same technology, I amclaiming that the internal heating (resistive losses) would be the same for the same output wattage.
However you need a certain amount of amps (which translates directly to torque) to turn the prop at the rpm you are now getting at full throttle with the 3s system. So that is going to put the 4s pack at some disadvantage. For the sake of argument, I'll assume the internal resistance of each cell is just inversely proportional to the capacity, so a single 1950 cell will be 4/3 the resistance of the 2600 cell (that's why I say we assume the same technology was used to make each cell). In addition there are 4 cells instead of the 3, so the 4s1950 pack has 16/3 R and the 3s2600 pack has 3R resistance (R=resistance of a single 2600mAHr cell).
Net result is that the 4s pack has 16/9 the resistance of the 3s pack (almost a factor of 2). This means that the voltage out of the 4s pack will have almost twice the voltage drop that the 3s pack experiences for the same current. Corollary is that it will also experience roughly twice the heating that the 3s pack experiences.
Another fact is that with the extra cell, you probably will get (at least initially) more than the ~55A out of the 4s pack at full throttle, and then of course the heating is worse. Since heat left in the battery is energy not supplied to the motor, and you say that the 3s2600 currently just has enough, my guess is that you may not have enough with a 4s1950 pack to finish the entire flight. In any event I don't see any advantage
One way around this is to go to newer technology batteries----These days I see 35 and 40C batteries. From my limited understanding of lipo's, the higher C represents a lower internal battery resistance (also may represent marketing hype!). I also notice that higher C tends to be higher weight too so you would always have to take that into account.
The lower internal resistance means that you will have a higher voltage output to the motor, which translates to more watts into the motor, and presumably to a higher initial speed. Of course you need to be careful that you don't overburn at high speed leaving too little in the pack for the full flight.
So readjusting for what you want--I guess you want more at less weight. Anything you can do to lower the battery internal resistance per cell means you have more to give to the motor--assuming the weight doesn't change. I have heard some really good things about some of the newer battery technologies--from people who seem to know and also have made independent measurements.
Here is a link to some of the testing being done with the newer cells (this is on RCGroups). I think the point from the test was that here are 4 manufacturers who have some pretty good cells.
http://www.rcgroups.com/forums/showthread.php?t=1087837These were 3s2200 cells, so to get the capacity you need you would need a 4s pack, and a motor with a kV ~3/4 of your current setup. However they probably have higher capacity packs too.
So I still claim that there is no inherent advantage (like in efficiency) to go up in Voltage. However if you simply want more performance with the same kV motor, then more voltage will force more current through the motor, but then you will need to carry more weight (=fuel)!
Hope this was of some help. I didn't really answer your question. I think to do that we would need some actual cell resistances and try to calculate how much energy is left in heating up the battery, and then add capacity to take that into account.