Have you ever studied a detailed procedure for heating up a large sodium valve from the solid sodium phase to the liquid? If not I highly recommend doing so. It's truly remarkable. Molten salt systems are the same level of complexity, but much more radioactive in the primary system. Remote maintenance of this stuff is totally doable, but it sure as hell is exotic. Rickover's characterization of sodium systems stands to this day.
Traditional sodium systems with void-swelling resistant structural materials may not go to high enough temperatures to strongly justify the tech development needed for a sCO2 balance of plant in the short term. Sodium-water steam generators with leak detection have worked well-enough to build a few more of them while sCO2 turbomachinery gets figured out at the 100s of MW scale.
Molten sodium is easy to pump, non-corrosive to many metals including steel, and a good heat transfer medium. The chemical industry uses it routinely. In a nuclear reactor, it would also chemically sequester a particularly annoying fission product (I-131) in case of a fuel element leaking. Unpressurized, it is good up to 800C before it boils.
By contrast, water at 300C will corrode most steels. At higher temperature, it will corrode zircalloy, forming hydrogen. And it really wants to be a gas, hence the giant containment buildings around LWR.
What did you say? "But sodium explodes in contact with water?" That's easy to solve, just keep the water out. A sodium cooled reactor should be coupled to a supercritical CO2 turbine instead of a steam turbine. That removes the problem of leaking and exploding heat exchangers.
(Footnote: I like molten salts better than sodium. But sodium is still better that water.)