You're right, I was oversimplifying. An aerial or submerged launch platform, then.
> the tendency is for a vessel to contract as it sinks and expand as it rises, leading to a runaway buoyancy shift (increasingly negative with depth, increasingly positive with height)
This is inherent to the cloud city design. Rockets would be a subclass of buoyancy risks, eclipsed entirely by atmospherics.
> Rocket launch from an inhabited floating atmospheric platform would require accumulation of large stores of fuel (the Tsiolkovsky rocket equation also works against you), as well as presenting various risks associated with enormous barely-contained explosions
This is a fair criticism. It's also solved by having offboard propellant storage and launch platforms.
> risks are immense, and hand-waving them away is disingenuous to the extreme
Didn't mean to suggest it isn't risky. Just that the risks from the rocket launch component are dwarfed by many, many others, and to the extent there are risks here, they are ones we've already solved on Earth in analogous contexts. (Maintaining buoyancy isn't remotely the main problem with launching rockets from high-altitude blimps.)