Just closing the loop: I received the EX-35R, and I measured it's idle current. If powered at 3V, it does indeed seem to have somewhere between 3uA and 4uA current (spec sheet says 3.5uA current). If powered at 9v, it has 8uA current, at least by my measurements, which were made using a Fluke 87V in combination with Dave Jones's uCurrent Gold. Even when activated following a confirmed detection (confirmed by pulse count that is), it didn't consume more than about 50-60uA at its peak, and the activation doesn't last longer than 5 seconds. So, it does seem very energy efficient.
Also, there's ample space in the back compartment for batteries and a Moteino. It could easily accomodate 3x AA. It might even accomodate c-cells, though I haven't tested that idea yet. Anyhow, there's a *lot* of space.
It's an open question as to how well it would hold up if exposed to rain. It doesn't look watertight to me, and I'm pretty sure heavy rain might get into it. I think it's probably meant for indoor use, but maybe (?) it would do OK in some covered outdoor locations where it's not getting heavily soaked. Or maybe you could just put it into some kind of watertight enclosure that can pass IR. Without making it ugly by covering all the edges in silicone, I imagine there should be some way to make it work outdoors. On the other hand, maybe if the power source really could last 10 years or longer, I could just seal it up with no intention of ever opening it again...
The one drawback is that it seems to take about 2.5 seconds to trigger after the initial detection, whether it be in 2 pulse count or 4 pulse count mode. It's much faster in "walk test" mode, which doesn't pulse count, but that consumes 9ma (how much of that is from the red LED I don't know) and obviously doesn't have the 2 minute wait for activity to cease before retriggering that the pulse modes have.