The Mac Pro, the Mac mini and the Thunderbolt Display beg to differ.
- The iOS Map disaster never happened.
- iPhone 4 users did simply hold it wrong.
- iPhone 6plus bending problems were only due to unrealistic conditions (and Apple switched to a stronger aluminum alloy and changed internal structure just for fun).
- The constant problems with discrete GPU's (MBP) and displays (iMac, MBP) in recent years were just made up by the press.
- iPhone 5's never had problems with nicks and dents in their chamfered edges.
- The generally amateurish switch from iOS 6 to 7 was working as intended.
- iOS9.3 did not fail to open links in Mail and Safari for many users.
And the list is not even complete. Yeah, Apple's QC is really up to snuff.
Besides the fact that Apple likes to ignore and sit out problems for a long time, before possibly(!) eventually acknowledging epidemic failure liability, their cash pile would allow them to survive pretty long before the problems would start to become visible externally.
IIRC they changed their service policies so that you now don't get a new replacement device for your defective iPhone anymore, but instead have to accept a refurbished device. Just a small change and seemingly uninteresting, but I find it pretty telling.
You never worked in QC and/or Software development, did you?
Not necessarily going to, but increasingly probable with increasing component combinations. In theory your approach sounds all nice and cozy, but unfortunately reality is very different.
Just for starters: Having those three notebook models, you need to test every software change against each of these models. Now multiply those three models with the possible component combinations and you reach three-digit variation numbers very easily, making driver and general software development and testing and QC very complex and error-prone.
Even within one model, Apple uses multi-sourcing for several reasons (production numbers, negotiation wiggle room etc.). So you could have different models of e.g. screen or SSD within two MBP's from the same model year. And when deadlines and budgets roll in, you _will_ make compromises in testing and QC - otherwise you'd never get any product or update through the door. The more variants you offer, the bigger the compromises.
And the above is not even considering the high dependency on external Software (drivers, compilers etc. etc.) and the changes made there, which might break compatibility with your current software. Sometimes also hardware components get changed for cost optimization or bug fixing, which results in yet more possible combinations. Then you'd have to roll out a fix that considers the situation both before and after that change. And you'd have to test this prior to rollout. And while you're testing, another component receives an update that interferes with the solution you cleverly worked out and extensively tested before. So it's back to the drawing table ...