Everything I can find about term 'metacognitive strategy' or 'metacognition' is related to teaching or learning strategies, but I wonder what does it actually mean in terms of cognitive science.
I don't think there's a single meaning for this term, mainly because "cognitive science" is pretty broad and inclusive. So people working in different sub-fields of cognition use it to refer to different things, depending on what they're working on. A couple examples outside of learning and teaching might be confidence judgements (how likely is it that your last response was correct? etc) in memory or categorization studies, or things like 'planning to plan' in resource-rational accounts of behavior, where you might want to work out how much time you can afford to spend making a decision. These are pretty different areas, but they are all legit ways to use the word 'metacognition', as are the learning and teaching examples you mention. They're all 'thinking about thinking' in some sense.
Actually if it comes to that I don't think there's a single clean and universally accepted definition for 'cognition' either. You recognize it when you see it, it's just a vibe sort of thing. Or, to put it more scientifically, 'cognitive' is a fuzzy category ala Rosch.
Confidence/error tracking example: https://royalsocietypublishing.org/doi/pdf/10.1098/rstb.2011.0416
resource-rational example: https://arxiv.org/pdf/1801.09848.pdf