Bummer - my cells are not here yet!
On the web site for A123, they show a max charge of about 4v and a max discharge of 2v per cell (listed in the temperature performance chart and the charge/discharge cyle chart). The charts definately show the voltage spike that you mentioned. BTW: the charts are on the performance and safety pages of the web site.
Given the 2200 mah that you have measured, the web site indicates that the cell should be capable of 100C burst, 80C surge and 30C continuous..... 220AMPS burst, 160AMPS surge and 66Amps continuous, per cell. If you are testing with standard lion settings, I think that you might get a bit more power using a manual voltage cutoff. Althought the charge cycle chart is a bit difficult to see, it appears that a max of 3.5v and a min of 2v is normal. But there is not much capacity below 2.5v. The chart also shows a peak of 40 deg C.(104F) for the tested charge/discharge that is similar to your tests. I noticed that the voltage starts moving up quickly near the end of the charging. I think that the mfg are cutting the charge off at 4v.
The difference is that even with extreme temperature spikes that would kill normal lithium cells, the carbon encapsulation prevents cell damage.
For Weight comparison - a Thunderpower 2S 2100 (20C) pack is listed on hobby-lobby at 3.2oz. That gives 44.8 grams for the cell and half the package and connector. I have no idea of the package and connector weight... maybe 8-12 grams???? That would give an estimate of 33 grams/cell versus the 70g that you measured. This is somewhat disappointing. However, the website shows a123 cell to be twice the energy density of Lion cells.... They don't really compare them to lipo cells. They show weight to discharge at 1500w as .9lbs (14.4oz). At 70g, that is 6 cells. A Polyquest20 pack (3S 3700) weighs 10.4 oz to output the same power.
To that end - these seem to be safer cells, but a bit heavier than Li Poly cells.