They actually are working on their own Qt3/Qt4 compatibility layer (probably the main incentive being not to have to package Qt3).
But from what I hear it is too straigtforward (a passtrhough for Qt3 calls and some just-in-time workarounds for Qt4 calls). It's probably the best they can do with their current manpower, since it allows existing apps to run with Qt4 with little/no modifications, but as it stands it is less performant than Qt3.
A proper port to Qt4 requires some refactoring, and that in turn requires a lot of manpower. Maybe they can make do with the compatibility layer while they slowly port apps one at a time, but a slowdown in the meantime is a tough sell, especially for a project like that which attracts low-speced machines.