Recently we had a few linting regressions, such as some dead code that was only dead with the default features but not --all-features, and issues with FFI tests falling out of sync (due to the unrelated test container issue).

It would be good to stay on top of this, and the cargo-all-features crate provides tooling to do this, cargo build-all-features finds many more bits of dead code for example, that we should probably be linting.

However, this doesn't work because cargo build -p payjoin --no-default-features deliberately fails for the payjoin crate, which requires at least v1 or v2 to compile.

That being said, it might not be desirable in the long run because the runtime for this doubles with each additional feature.