I think the only way to cut through all the confusion is to recognize "facts" as "predictions" that can be reliably tested - where the standard of what's "reliable" is in the eye of each individual person.

That might seem too subjective, but I think the necessity of pragmatism naturally prevents the subjectivity from getting out of hand. It's like your first quote: "Science is not powerful because it is true, but true because it is powerful."

It's the "predictability" of facts that make them pragmatic, which makes them powerful, and which makes people tend to agree on them, as suggested by this other quote: "What convinces masses are not facts, and not even invented facts, but only the consistency of the system of which they are presumably part."

I can predict that if I leave water boiling, it will eventually all disappear. That's a fact, because it's a prediction I can verify. Mathematical statements, like "1+2=3" are "factual" for the same reason, namely, that they follow a pattern that we can learn to "predict".

We can bring the same strategy of "predictability" when it comes to interpreting textual facts.

If we find someone's notebook where they are plotting a murder or assassination, we can "predict" that they are a threat, and should act on that. It's not about "knowing" what's in their heart of hearts and "between them and god" - it's just about us making useful predictions that will bring about the outcomes we desire.

Every problem / challenge of establishing facts that you've presented in this article can be settled in this way - simply by recognizing "facts" as "predictions", and recognizing all acts of "logical reasoning" simply as acts of accurate prediction.