Cells in series do have this big advantage in efficiency and lower amps, as TA explained well. The downside is complexity on the driver electronics side if buck or boost circuitry is needed. We are getting better in our open source BLF drivers for doing them, but it's a more advanced design, requires more parts, costs, etc., more driver PCB space, etc.
It will be our future though - I believe this is where it's all going, just like electronic switches. It's already there in most higher end commercial flashlights. Mostly for more power, you need more space, plus there's more cost.
I was seriously considering doing an in depth comparison of a Q8 (parallel) against a Convoy L6 (series) to weigh all the pros/cons:
efficiency, runtimes, output, throw, form factor, size/weight, costs, mod-ability, safety, etc.
I'd also like to do a straight parallel SRK light comparison to a modded series SRK. Apples to apples sort of thing - same LED's, same thermally, same cells, and see the real #'s.
This, of course, would be a huge amt of work to do, but would be interesting. I know if you have to wire up say, 4 to 10 LED's in parallel on separate MCPCB's in an SRK style light, it's a real pain - series is easier, and a big plus is you don't need to use as heavy wires.