Not supporting the "recovery move" motion, IMHO, because it is not consistent with actual Recovery Moves, or even HP Draining Attacks, for the matter (Protective/Evasive moves, anyone?)
To me, this move has the exact same issue as Endeavour - they fulfill a niche situation where using them would be far more beneficial in the long run than spamming 20+ damage attacks. You need to actually be in a serious HP disadvantage (somewhere around 38 to 50) compared to your opponent for Pain Split to be actually effective, whether by chance or by design. If we're worried about Pain Split ruining the "Good Stats and 20++ Damage Attacks per Action" trend in battles, then I'd suggest sky-rocketing the EN cost.
Why? Simple. Under the said "trend", you'd never exhaust a Pokemon's EN before it is KO'ed. So let's provide a situation where it becomes a HP-versus-EN conundrum for the player. By all means, make it so that Pain Split costs more if not as much EN as a good combo. I can't delve into the maths right now, but this is my proposal for your scrutiny :)
EDIT: Reposting this for thread revival. Here's my proposal:
Energy Cost Formula: 6 + (|Opponents HP - User's HP|) / 3 6 + (|Opponents HP - User's HP|) / 2
Basically, if we near-doubled the EN cost, I think it would serve as a deterrent for Pain Split spam, especially in longer matches where conservation of HP and EN is important. Under the proposed changes, at the best circumstances (exactly 50 HP differential) you'd spend 31 energy just to regain the HP footing on your opponent. Of course, if peeps still think it's not severe enough, we can even remove the factor to raise the EN cost. Heck, add in a multiplier if we want to go all the way...