We absolutely should not. In fact, we should go further and state that Cargo.lock should never be committed at all. Committing Cargo.lock, regardless of crate type, makes bisection and reproducible builds marginally easier, but makes all other uses of the source repo significantly less ergonomic.
(I have said this before, e.g. in Feedback on `cargo-upgrade` to prepare it for merging - #122 by zackw. I promise to post an actual explanation for my position ... probably as a new thread ... sometime this week.)