Does Astronomy/Cosmology rate an exception?
They have a wealth of observations and models and conclusions, but test?
A test does can amount to things, for instance, the measure of light bending via gravity. There is no manipulation required. The only part of physics/cosmology that might be untestable is some string theories, and there is objection to those on that basis, as they can describe many kinds of universe just by adjusting coefficients.
Things like astronomy can be tested, i.e. "the orbit will go there". Did it? Yes? Good, then the theory is acceptable. Continued success leads to an accepted theory, until, of course, error is discovered, and then refinement is required.
A good example is the evolution from Newtonian Mechanics to simple Einsteining mechanics, to full relativistic mechanics, which so far is batting 100%.
Likewise, measures of things like absorption lines in spectra, etc, can be used, carefully, to do some very surprisingly accurate testing.
So, a test can be an observation of a prediction or the like.