My point in a nutshell:
Good enough is usually an excuse for vague and ill-defined practice. If you don't have a well-defined "good enough" you probably don't have a mature process. If you don't have a mature process, you probably shouldn't be writing critical code. Hence my original comment that it's not a good mindset for high-risk applications.
Well defined "good enough" often looks like a standard. Those standards should be risk based so that one person's biases don't result in a different level of risk mitigation than another person's. That risk is what contextualizes what is "good enough". I'm sure if you asked the Boeing managers, they felt their CST-100 software was "good enough", but the relevant safety standards say it wasn't "good enough". Since both sides can use the term, it makes the term somewhat useless. Like you say, there is no singular "good enough" so the question becomes: Good enough...for what? Good enough to meet schedule, or good enough to not risk crashing into the ISS? My main issue is that "good enough" often means "undefined". When people say "good enough", I've found it often means "we don't know what we need, but I'm sure we'll know it when we get there." I think that can be a bad approach to software development when the risks are high because it opens one up to cognitive biases that lead to subjective and poor decision making.
If you're saying "good enough" is precisely defined and based on risk, then I agree. But that is not how I've ever seen the term used in practice. It's almost always a nebulous term which means you've only vaguely defined the risk. Poorly defined, subjective judgement belongs more to art than engineering, especially not safety-critical engineering. "clunky, cobbled together, expermental or just plain stupid" as the OP said, just doesn't cut it on critical applications, even if the developer claims it's "good enough" and doesn't strike me as a professional mindset.