Normally, for correct operation, put in parallel identical semiconductors. In the above case, the current margin is sufficient to ensure proper operation.
LED does not work in this case in commutation, only impediment is the temperature dissipation.
If that driver out a voltage PWM type complications can occur.
I think it goes. COB LED majority are in series and in parallel.
And even if you are switching, if there is a good cooling, 2.1A for a few microseconds is not fatal.

Naoki, it wasn't my choice to create a parallel circuit. However, with 14 diodes each drawing 15 V, I'd need a driver that outputs over 200 VDC at 300 mA for an all-serial circuit. Not likely!! So I got more reasonably spec'ed drivers. No brand name on the name plate, but no complaints. It runs at about 50 deg in operation. I still haven't bought a Kill-a-Watt meter, it's on my task list.

I haven't noticed any diodes running brighter than the rest, but measuring current in individual branches is not feasible. My light meter hasn't detected any difference in brightness along the wall.

Now, taking underdriving a step further, what would happen if I doubled the number of LEDs? Would they work at all, or do they need some minimum current in order to light up? If underdriving increases efficiency, I should get more total light output than with 14 LEDs, right?

Thanks Nexogen, I didn't think that temporarily overdriving the LEDs was very harmful. As I mentioned above, practical driver specifications dictate that the diodes are wired in parallel/series rather than only series.

Well, if it works, then you don't have to worry about the current hogging due to the Vf variation!

You'll almost always gain efficiency if you under drive them with more LEDs (by doubling the parallel strings). The amount of efficiency you gain will depend on the specification of LEDs. For example, the Cree CXA gains a lot (compared to Vero) by under driving. If you have the data sheet, you can calculate the payback period. But in your case, you already have a set amount of energy consumption determined by the driver (i.e. Vf * 2 * 2.1A, if Vf = 15, then 63W). Well, the actual consumption should go down a little bit because Vf goes down a little bit with lower current. But it is probably not so big difference. So it is a bit weird to think about "payback" period.

Without the data sheet, it is difficult to say if it is worth doubling the LEDs. But the total output could increase 5-15% (20% or more increase is unlikely). You can go pretty low current. With Vero 29, 2.1A is the nominal. You can run it at 350mA (17% of nominal). I don't know how the efficiency behaves at the extremely low current, though.

Here is something a bit related, but it is probably not a big factor. Most CC drivers have the maximum AC-to-DC conversion efficiency when the output voltage is closer to the maximum voltage of the driver. If each LED is getting a lower current (i.e. doubling the number of diodes), Vf goes down (so lower total output voltage). But the penalty here is likely to be less than a few percent in AC-DC conversion efficiency.

Sorry, maybe this is way too much nit-picky details...

You already have the driver, but a person in another forum tested the following low current, high voltage drivers (for serial connection), and they are pretty efficient (90% or so):

$8.12, 54-120V DC out, can drive about 5 Vero 10 or about 7 of yours:
$8.12 18-36*1W 85-277V High Power LED Driver - output voltage 54-120V at FastTech - Worldwide Free Shipping

$3, 39-75V DC out, can drive around 3 Vero 10, or probably 4 of yours:
Free shipping 13 21W high power led driver 13 21x1w led lamp transformer 300mA driver for LED DIY light transformers AC85 265v-in Lighting Transformers from Lights & Lighting on Aliexpress.com | Alibaba Group

They are around 280mA.