Do you think the engineers at the "original" Lockheed skunkworks would allow tech debt to creep in on the SR-71? They took measured and balanced shortcuts (that thing leaks fuel like a sieve when on the ground, they optimised it for supersonic flight), but they didn't take less care about the technology they were building.
Rather than accepting tech debt, the way to think about it to my mind is "we're going to leave out some functional and non-functional stuff we don't need", but those things are rarely tests or maintainable code. They're more like actual user functionality. What's there is tight, sleek and focused, but it's finished.
I took the context to be software skunk works. To give some examples, when SMF and ZFS integrated into Solaris' OS/Net core there were many unfinished things and bugs and many devs outside those teams had to jump in and help. I'm not sure if that's tech debt or what, but definitely they moved fast and broke things, yet they also delivered incredible value.
Would that be acceptable if building something like the SR-71? I want to say "no, no way", but then look at SpaceX's approach with Starship. You wouldn't put humans on such a thing until it's done, but if there's bugs to shake out, so be it.
And in fact the SR-71 did have bugs that needed shaking out, and one pilot died because of those.
