Doesn't matter for majority of the playerbase. External testing lasts only for a short time to get a glimpse of how the fight works, and to check if there are any gamebreaking bugs with it. A foobared line of code can make the diffrence between playable and unplayable boss. For example boss has an aura that makes it take 100% addidtional damage in a certain phase, like Gorefiend. Now if someone slipped a finger and made it 1000% instead of intended 100, the fight would be utterly broken.
Testing avoids situations like this.
The only instance where I disagree with Blizzard's decision was when people get banned for unintentional "exploitation" on world first races. The most notable example of this is Lich King saronite bombs. Fortunately, this happens extremely rarely, in fact that Lich King kill was the only major incident. Abusing LFR in Dragon Soul was kinda intentional cos they knew how many bosses were meant to be available on that particular date.
Today's raids are saturated with mechanics (too many mechanics for my taste tbh, IMO there should be much less stuff going around, but the stuff that IS going around should be meaningful and harshly punishing if failed). They need to ensure all this mess works correctly.