I actually knew a rogue once that took Cold Blood over Surprise attacks....then again he also used a 2.6 speed offhand...

As was mentioned previously, each additional, separate haste effect (S&D, BF, berserking, mongoose, weapon procs, haste rating) improves the effect of other concurrent haste effects and should therefore be used simultaneously. The background mechanics of what's going on is that each haste is hasting the other hastes in addition to the normal white damage, so the more hastes there are to haste each other the more benefit you get by simultaneous activation than by separate activation. This also means that each separate haste effect can be looked at as a separate multiplier on your dps, that is, when looking at the dps gain going from no BF to BF (for example) you don't have to consider whether S&D is up; the percent increase is the same in either case.

However, since all haste ratings are lumped together into a Haste From Haste Rating modifier, it's worth noting that in the absence of any other haste effects there is no difference in activating rating-based haste effects simultaneously or separately. 100% haste from rating gives double the attack speed, a 100% dps benefit. 200% haste from rating gives tripes the attack speed, a 200% dps benefit. In the absence of other cooldowns (haste or otherwise) to coordinate with, there's no difference in overall effect between having 200% for 10 seconds or 100% for 20 seconds. (these values are eggregiously impossible and chosen merely for illustrative values, head-math is easier with integers)

Summary: You should bother coordinating haste effects that are up only some of the time (ie not quiver or S&D) only if at least one of them is not rating-based. For raiders this generally means warriors that are neither trolls nor have the mongoose enchant (and honestly the mongoose effect may not be worth the effort), although they would still want to coordinate all of their cooldowns with deathwish for maximum effect.
As people mentioned, you shouldn't wait too long for a proc to happen, where too long means you will get one less opportunity to use it during the fight. If cooldowns don't exactly line up, you're gonna have to crunch numbers hardcore.

As far as melee haste, casting haste, and the recent change:
First, melee haste rating was (probably) a wee bit OP, as in it tended to give more dps increase per point than any other rating, having a 7rating->1% conversion at level 60, with the next-cheapest stat being melee hit at 10. This is a rather large disparity, since one percent rating gives an equal dps increase for both, and they both only effect white damage (when hit-capped on specials). As far as I can tell, the only situation where hit% does more damage than haste% is if you're not hit-capped, and yet haste was far cheaper. They're now aligned in cost to 10 at level 60, 15.7 at 70.
(I like looking at the 60 conversions for comparison purposes. Once again, integers.)

I think the original intent in having the ratings priced as they were was that caster haste would affect all of their dps, while melee would only gain an increase on their white damage. What we've seen instead is that with combat potency, hunter shot weaving, and the rage mechanics, melee get an increase to more than just their white damage, whereas casters with their instant-casts and mana consumption tend not to get the full listed improvement. What we have now is that, in most raid situations, 1% melee haste gives up to a 1% net dps increase (usually less), and 1% caster haste gives at most a 1% dps increase (usually less). The points values are now actually representing this.