1. Is there any way to anticipate the isolation distances needed between the primary and secondary circuits?
Its just distances given in the standards.
In a transformer you need often to use triple insulated wire for sec, and say enamelled coper wire for pri...then you pass standards ok.
You look into it very deeply......in truth, you only need to think of the regulation hipot test..that test device will put the voltage that you speak of between pri and sec, and your insulation must not break down in that situation.
You ask a very deep question, and i suppose a freely floating secondary coudl float up to some arbitrary voltage above earth...and who can say how high it could go....the primary is the mains and is all ultimately earth referenced..........and i believe in theory, it is most likely that the secondary , if freely left to float, would tend to straddle earth potential....so if it was a 24v output, then probably it would come to rest at -12v to +12v W.R.T. earth.......but many say it would come to rest straddling half mains peak (either pos or neg).
But say if your secondary ground is connected to earth......then you could argue that the secondary circuit would not realistically be able to rise much above say +/-339V (W.R.T. earth).....under any circumstances...because even if a 6kv transient hits primary side L-N. then realistically, in a PFC'd supply, the cap bank would quench it down before it got near your txformer.
But yeah.....there is a capacitive divider goinng on in the trxformer, and that has some impact.