Woah Nellie! <who?> Everything I just said is screwy. I was thinking about assuming what it does in one second or one year is done in one hour, which is not sane at all, of course. This question was scratching at my head all day; CARNAL1, I know it's not what you meant to ask, and I'm not jabbing ribs, but it's still fun to try to answer the question correctly -- why not. Considering unexpected questions burns new neural pathways, and unexpected results are sometimes useful, or at least interesting to consider....
Drawing 670mA for one second is not 670mAH, it's 670mAS (milliampere-seconds), which, when multiplied by 1HR/3600Secs is about 186nAH (nanoampere-hours). So what that really means is the charge in the battery was reduced in one second by 186nAH. But that's not a fair measurement of the power consumer--it didn't quit on its own in just one second. More fair: In one year, the battery's charge may be depleted, or nearly so, but not to zero by the flashlight as much as its own internal resistance, assuming the flashlight actually quits drawing current after a certain point prior to depletion; so, not all of that is attributed to the flashlight, then. Some of it. Connected to an endless power source (or an endless supply of little power sources), the mAH of a flashlight is infinity, or until it breaks! So, that question is really asking about the lifetime of the device.
What is the mAH of a resistive load? Connected to an unlimmited power supply, it's infinity mAH (or, uh, infinity AH), or until it breaks.... Connected to a battery, it's roughly the capacity of the battery.... until you connect it to another battery, that is. So, what you connect it to needs to not to play into the question "AH of a consumer" at all, then, because the question didn't mention the source. So that leaves it only to mean "what is the total charge that this consumer will use over its entire lifetime?"
Conclusion: I'd like an endless supply of little power sources.