Prototype bindings for Firefox Desktop platform (C++) consumers

[linabutler]

Once we have our ADR for platform consumers (#268), we’ll want to make a prototype. I don’t have many details to add here yet, since we don’t know how we’ll approach this—that’s what the ADR is for! 😁

I think we can reuse the core of #255. The RustBuffer serialization machinery is going to be useful here as well, except instead of serializing WebIDL-specific types like mozilla::dom::Nullable, dom::Record, and dom::Sequence, we can serialize mozilla::Maybe, mozilla::HashMap, and nsTArray, respectively. We’ll also need to generate some code for exposing functions, records, and objects to C++, too, like we do for Swift, Kotlin, and Python.

Good news: because we’re only generating a C++ API, and not any IDLs, I suspect we can land this in Desktop quicker than whatever we do for JS.

┆Issue is synchronized with this Jira Task
┆Issue Number: UNIFFI-18

[linabutler]

Let's leave this open, but backburner it for now. Kate and Ana would like to focus on front-end experiments first, and figure out a story for platform later. Since the Nimbus team doesn't have any platform consumers yet, we'd be on the hook for maintaining Nimbus, the C++ bindings, and the actual experiments themselves, which doesn't feel like the right thing.

#285 was helpful for figuring out how a pure-C++ API would look, but it's not worth our time to push on that since Nimbus doesn't need it right away.

[rfk]

Closing old Gecko issues and consolidating them under #991