There's a bit of false conclusion there, though. The number of heals you cast during the whole BG is not a related or significant number. Each individual cast is what matters. If I throw six heals on six people each six seconds apart, the fact that each one was .06 seconds faster does not matter. The only way haste would matter in that situation were to be if the heal lands so infinitesimally close to the moment the killing blow lands, that that .06 second somehow made the difference. And keep in mind, that is six pennies in a dollar, of a
second. Server latency is greater than that.
Haste would matter if you're chain casting on someone, I agree, but let's look at by how much. For starters, I'll take the total mana pool of my rdruid,
Spreelit. Her total mana pool is 600 and her combat regen is 39/5, or 7.8 mps. Regrowth has a 1.5sec cast time and costs 36 mana, or 24mps. 24-7.8=16.2. 600/16.2=37.03 so just chain-casting Regrowth, an rdruid will go OOM in 37 seconds. So any discussion involving chain casting longer than 37 seconds is moot. That OOM time is faster if you're hasted (everything costs the same but you burn through it faster) so for the sake of argument let's talk about chain-casting on a party member for a flat 30 seconds. How many heals can you get off on one target, chain casting, in 30 seconds, with or without haste? At a casting speed of 1.5 seconds each, you can get 30/1.5=20 casts off. With haste, you get 30/1.44=20.833 casts off.
So... even considering
no latency and perfect casts, and considering you don't have to do
anything, in a BG, but stand perfectly still and cast, and your target stays alive, for
30 full seconds--in a battleground, mind you---by the time you run
completely out of mana you
still will not have gained enough speed to pull off even
one additional cast. If you had Lifebloom, and were a troll and had Berzerking up, then you
might have just enough extra speed to pull of ONE additional heal,
in thirty seconds of doing nothing but standing still and healing and perfectly compensating for lag on every cast.
I wasn't sure what the numbers would produce when I started calculating this, but that's sort of hard to argue with. Anecdotal "it feels faster" arguments don't really matter in the face of numbers. If anyone sees a problem with my mathematics, feel free to point it out, but it appears that even with all the haste we can get, you would have to concoct an idealized, laboratory-perfect scenario with every single source of haste available to us in order to make even one tick of difference.