I believe quirks would be added as they are uncovered. It'd also become a form of documentation. Emulation doesn't need to be precise to be valuable - if we can test it against an emulator on a commodity machine before testing it on metal on a somewhat busy machine, it's still a win - we don't need to waste time on a real multimillion zSeries or Cray if it crashes the emulator.
That way we would be able to create integration test coverage of 100% while still being able to panic on real hardware, until all messy behaviour is implemented. It's like writing unit-test kind of integration tests which then fail when provisioned :)
Passing tests and then failing when provisioned is still better than just failing when provisioned. At least you know that it should work (and maybe does, on different variations of the hardware).
I believe quirks would be added as they are uncovered. It'd also become a form of documentation. Emulation doesn't need to be precise to be valuable - if we can test it against an emulator on a commodity machine before testing it on metal on a somewhat busy machine, it's still a win - we don't need to waste time on a real multimillion zSeries or Cray if it crashes the emulator.