Perhaps on a straight, but in the bends or slopes you'd still need some sort of control. I'm visualising the kind of oscillations that luge or bobsled riders get in the winter olympics when they come out of the corners.
High speed and low friction could mean a lot of potential lateral movement that needs to be controlled somehow. Maybe releasing some of the stored compressed gas in controlled bursts like used in various space endeavours. That would have the added benefit of off-loading some of the potentially dangerous steam.
I've never operated a bobsled, but if they're like other machines I've operated with physical controls that have mass or that move in response to things that have mass, then the oscillations are caused by the driver. Which makes sense because the better bobsled drivers are observed not to suffer from these oscillations as often as their competitors. (In fact we might say they are "dampers" rather than "drivers".)
In the absence of any controls, we would expect the train to climb the side of the tube on a corner, without ever reaching a "horizontal" orientation. When exiting a corner, perhaps there would be a pendulum situation, but I'm thinking that a train that wasn't round in cross-section would damp that pretty quickly. If not, the system generates compressed air. That can be released through nozzles that are oriented and controlled to dampen the rolling.
But you're right, of course. A bottom-heavy train is just a more effective pendulum.
High speed and low friction could mean a lot of potential lateral movement that needs to be controlled somehow. Maybe releasing some of the stored compressed gas in controlled bursts like used in various space endeavours. That would have the added benefit of off-loading some of the potentially dangerous steam.