Your hypothetical mastering engineer, who masters on a single flawed speaker and, what, hasn’t even bothered to zero out the frequency response before analysis, is ludicrous. This hypothetical person/situation is, as you say, “poppycock”.
A circular, self-referential straw man argument—precisely like the one kenjit set up with this caricature of a completely unprofessional mastering engineer—is a worthless rhetorical device. If your goal is to arrive at truth, this is a disingenuous way of going about it.
In reality, a professional mastering engineer reviews program material across numerous grades and styles of speakers; from mobile devices to planar magnetic/electrostatic, to massive vintage JBLs or Altecs, and even gigantic audiophile multiway towers like Dynaudios and Wilsons.
The situation kenjit posits, a poorly mastered recording made by an unprofessional engineer, is useless because such a poor recording will only sound correct on a system with the same flaws as the original one that the recording was mastered on.
This is precisely why a professional mastering engineer first zeros a mastering system to a reference standard before going to work. It is also why this entire thread is pointless.