These questions belong to the "business side". That means, in the XP parlance,
the Onsite Customer owns the backlogged priorities.
So if the OC must ask the programmers (or, better, the marketects on the
business side) to perform some research, they drop this request into the backlog
as a "small research project".
An important rule of XP is, if the business requirements for a given project are
not yet clear, that is a major sign you should not do the project. You should
work to add features that the business side is literally dreaming about - they
must want them so bad they can describe their vision for them in detail.
However, a conflicting rule of life, in general, is you should start early if
you'd like to end early. So that research project should get as high a priority
as possible!