Why hold conferences in a resort town?
I cannot talk about general customs, but for whatever it’s worth, a major conference in my field is held in a ski resort, off-season. This has two advantages:
This is what the resort does to survive during the warm seasons. Therefore I presume that it may be a relatively cheap option.
Except for hiking, there is nothing to do there. With conferences held in cities, it often happens that attendants take some time off to do some sightseeing or similar. Here, getting to the next (touristically boring) city requires you to rent a car and takes one hour. Therefore, the conference gets more attention.
This is an example of a social norm, which is a kind of large-scale Nash equilibrium. Namely, there is no special justification for why academics benefit from the perk of being able to travel to attractive tourist destinations for their professional meetings, other than the fact that this is the norm that has developed historically, and once the norm has developed, it is stable against disruption, since no particular player in the game (in the sense of game theory) that is academia has an incentive to disrupt it. The only people who might object are funding agencies who may prefer if conferences were organized in drab, cheap locations, but if ever any such agency were to propose to cut off funding for conferences held in interesting places, the people being funded would cry out that this would hurt the competitiveness of their conferences relative to other similar conferences being organized by people funded by other more permissive funding agencies, and the reform would be scrapped. This is precisely the general dynamic at work that helps maintain many Nash equilibria/social norms.
That being said, the tradition of having conferences in nice places also benefits scientists, and therefore science, by allowing them to do their work in a pleasant environment that is conducive to stimulating creativity. Of course, people in other industries would also enjoy these sorts of conditions, but they can't all arrange them for themselves, so perhaps the right question to ask is not why academia can do it, but why other industries can't.
Another thought is that holding conferences in attractive tourist destinations also provides an economic stimulus of sorts to the travel and tourism industries of many countries, so may not be a bad thing, and by providing employment to multiple economic sectors it attracts political support that again makes it difficult to eliminate this particular "market inefficiency" (such that it is).
That differs a lot by discipline: in mine (sociology) the annual conferences of the different associations/societies/sections/etc. are typically hosted each year by a different university, and the conference takes place not in a resort or hotel but on campus. Only the really large conferences have to get creative in order deal with the large number of participants.