Actually neither of those scenarios seems right to me. At some point scientists should generate a falsifiable hypothesis and then test to falsify that hypothesis.
You are correct that this step seems to be omitted in much popular writing about science.
This is a sciency-sounding position that actually misrepresents science. Since scientists are not gods, just as with macroeconomics and astrophysics and paleontology, climate science cannot readily “test to falsify a hypothesis.” We are restricted to observing the universe and hoping to come across useful data we can compare and contrast. Thus to imply that the inability to regularly cleanly isolate and test variables in these disciplines is unscientific is itself a disservice to science.
No, that pretty much is the definition of modern (Popperian) science. At least the sort of science that we should require in order to justify rewiring trillion-dollar economies.
This "astronomy makes no predictions" meme is goofy. Halley got a comet named after him by predicting the year of its return. That was a completely falsifiable hypothesis. If you like something a bit more current, there are plenty of falsifiable hypotheses concerning e.g. neutrino mass that various underground/under-ice detectors will test in the next few years.
Such hypotheses are not inherently impossible for climate science.
You are correct that this step seems to be omitted in much popular writing about science.