Well said AJay. I am from the development side myself so I know exactly what you are talking about regarding testing.
From experience QA is there to ensure there are no game breaking bugs essentially and that you will pass through First Party hopefully on the first attempt. You would be surprised how much development time is spent on trying to accommodate first party checklists and guidelines, as a developer you'd rather not spend time on those sorts of things but it is what stops you from shipping a game ultimately not the fact that a powerup is too powerful.
Game balancing is a challenge and you can't catch everything, not to mention changing one thing has an impact on something potentially unrelated elsewhere in the game. The final build of a game is more likely to only see 100 or maybe 200 hours of testing done to it prior to submission. At this point they are likely fixing one or 2 bugs so you generally test those fixes and then smoke test the game at that point to ensure it didn't break something. It all depends on your timelines and ship dates.
On an unrelated subject the issue they had to fix server side regarding being unable to obtain more scores after playing 1000 songs i think it was? That is evidence right there that you simply cannot test that stuff in a basic testing environment. They should have populated a testers leaderboard with dummy data for every song in the library to test this but that is hindsight and I'm sure HMX have learned from it, as they will learn from this.