Critical mass in a molten salt reactor

If the fuel concentration drops you can extract the salt from the solution as well perhaps using a chemical reaction that converts one of the components of the salt to gas or solid.

That safety reservoir isn't going to be a single big hollow tank. Instead it's going to be several smaller tanks lined with neutron absorbent material and designed to quench the reaction as fast as feasible.

Thorium is a non-issue, it can't go supercritical. The issue here is that we're talking about neutron-induced fission, and a Thorium nucleus that fissions will only release a single neutron. Since the some of those neutrons escape, it's necessarily subcritical.

Uranium will release 3 neutrons per fission, so this can cause chain reactions. But the critical mass depends on the geometry. A sphere is the worst case (lowest critical mass), while a flat disk of negligible thickness has a near-infinite critical mass. Virtually every neutron will escape the disk without hitting another neutron.

This still applies when the Uranium atoms are in a molten salt. And fluids can change their geometry rapidly. For instance, with the safety reservoir below the reactor, there's no need to drop the whole working mass. You can leave some in the reactor itself, and the fluid that does flow out will naturally form a disk.