The language it's written in has very little, almost nothing, to do with how efficient larger applications are.
This is almost entirely up to the design and day-to-day decisions of the developers. These almost always outweigh the efficiencies of the underlying languages themselves (within reason).
A single location of poor data access patterns could negate the aggregate performance gains of your entire application, as an example. A framework that prevents you from making simple mistakes and drives you towards more efficient patterns goes much further than the language is written in.
Between Rust, C#, Java, and Go you're essentially even on performance for large applications (with C# pushing ahead of the pack). What you are not even on is engineering efficiency, it's going to take considerably longer to build the same set of features in rust than any of the others listed. And the performance is likely the same, potentially even worse depending on the maturity of the ecosystem.
Rust is a great systems design language and a great language to choose when developing high efficiency libraries & frameworks for I/O and data processing. It's not really a great choice for application development due to how slow it is to actually get things done in.
I fully expect to see alternate backends written in more operationally efficient languages over the next decade that will catch up to the official Lemmy codebase, and potentially even replace it. It actually sounds like a super fun project, funding is always a problem though.