With an AAA-to-AA adapter in hand, the Eneloop AAA cells looked like this:

The glitch comes from a not-quite-seated cell, showing that a poor connection matters.
The package touts “up to 800 mA·h, 750 mA·h min”, with asterisks and superscripts leading to “Based on IEC 61951-2(7.3.2)“, access to which requires coughing up 281 bucks. So it goes.
A full charge made them happier:

The as-delivered 530 mA·h capacity represents 73% of the 725 mA·h after the first charge, so I suppose they’re more-or-less within the “Maintains up to 70% charge after 10 years of storage” claim. The 16-10
date code suggests they’re hot off the factory charger, so they must ship with somewhat less than a full charge.
Comparing the capacity in W·h makes more sense, because most devices (other than the Planet Bike blinky light these will go into, of course) use a boost converter to get a fixed voltage from the declining terminal voltage.
They arrived bearing just over 600 mW·h:

After charging, that went a bit over 850 mW·h :

Call it 71% of full capacity on arrival. Close enough.
The Planet Bike blinky will be somewhat dimmer with two NiMH cells delivering 2.3-ish V, compared with the initial 3-ish V from a pair of alkaline cells. I generally burn the alkalines down to 1.1 V apiece, so perhaps they’ll be Good Enough.
Now, if I were gutsy, I’d install a rechargeable lithium AAA cell, with a dummy pass-through adapter in the other cell socket, and run the blinky at 3.7 V. At least for a few moments, anyhow …
they must ship with somewhat less than a full charge
This is actually the case. Apparently the regulations (as of April 2016) require that lithium batteries be shipped with a 30% state of charge. I’m not sure quite why, but I’m guessing that this is considered the safest.
Probably to limit the energy/max current they can supply and keep the cell from deteriorating at the same time by not going flat. Eneloops are NiMH chemistry though, so LiIon reg. probably don’t apply. I’d peg it on diminishing returns, they give you the cell ready to use but save 20-30% on charging – kind of like smaller ink cartridges with new printer but less costly for the end user. Also, they can probably charge to 70-80% with higher charging speed then the last 20-30%, right?
If you want higher voltage, insert a 0.7V drop diode in the dummy cell and use 3.7 LiIon. You’ll piss away some energy, but then how much current does the blinky use anyway?
The right solution, of course, would be to build a buck regulator in the spare compartment, rewire the contacts a bit to make it work, then feel the warm glow of energy saving. [sigh]
Thought about it, but how long would it take for the energy savings to recoup the energy you’d spend making it? :)
LEGOs current motor architecture for Technic series features two sizes of battery holders (AA and AAA) but they also make the smaller one with integrated LiPoly pack. I played with the idea of making the LiIon version of the larger one to power bigger models, but I didn’t get further than back-of-the-envelope doodles due to size restrictions and size of off the shelf eBay converters. Making my own is beyond me… well at least beyond my will to do so :)
Evaluated on that basis, nothing around here has a finite payoff period!