The fellow in the article who ruled out #4 pre-emptively, saying it's as likely as talking with Santa Claus, manifests an anti-scientific prejudice.
He ought to have said that there are tests one could devise which, while not proving #4, could go a along way toward disproving #1, #2, and #3. If her brain does not show the physiological characteristics or measurable abnormal activity associated with hallucination; if the entity she's in contact with worships Jesus Christ our Lord and teaches nothing contrary to faith or morals; and if she has a reputation for truthfulness and passes a polygraph: as I said, it doesn't prove, but it vastly increases the probability of #4.
There are studies in which a subject has a magnetic field stimulated on the side of the brain (right side, IIRC) just above the right ear. The subject ‘feels the presence of beings around them, though the pre4sences are not visual.