rust-analyzer: Set a library's display_name when consolidating crate specs

[reiyw]

This PR addresses the issue #1032.

When there were multiple build targets sharing the same source, their crate specs were consolidated into one, and their display_name depended on the result of aquery. Since the display_name is actually a target name, it caused a mismatch between the display_name and the actual crate name referenced by the Rust source, leading to unresolved import errors.

To eliminate the possibility of the name mismatch, this PR changes to set a library's display_name when consolidating crate specs. This change resolved the unresolved import errors of Rust Analyzer in my reproduction repository of the issue.

[hlopko]

Hi @reiyw, thanks so much for working on this!

So, what do you think about instead of checking the crate type, we would check for presence of wrapped_crate_type - that would imply that the crate at hand is actually a wrapper like rust_test using crate attribute or potentially other rules that are not part of the rules_rust, but use the integratoin APIs of rules_rust. In case of a merge event, we pick the display name of the crate that does not have wrapped_crate_type.

For a merge event of rust_binary and rust_library, I agree that library should win, but is this actually a situation that happens in practice? I'd expect that rust_binary will have its root in main.rs file, and rust_library in lib.rs file, so there won't be a conflict. I'd be fine with discussing/solving this in a followup PR, since the rust_library/rust_test problem is far more pressing.

[hlopko]

Hmm, I changed my mind, let's go with the current PR approach.

[hlopko]

Thank you again!!