Enable sharded repodata for `repoquery search|depends` command

[travishathaway]

Checklist

• I added a descriptive title
• I searched open requests and couldn't find a duplicate

What is the idea?

We should add the new sharded repodata feature to the conda repoquery search|depends command in this repository.

Why is this needed?

For faster searches and feature parity with the rest of the codebase.

What should happen?

• Update conda_libmamba_solver/repoquery.py so that the LibMambaIndexHelper objects created there are passed an in_state (SolverInputState object). This in_state object needs to be aware of the current environment and the search term being passed via the command line arguments.
• Make sure that the code path for the whoneeds subcommand does not use sharded repodata. That needs the entire repodata to search all packages that rely on the provided pacakge name.
• Add tests to verify these searches are actually using sharded repodata fetching (use the http_server_shards fixture for this).

Additional Context

No response

[dholth]

For search we need a way to fetch individual shards without going through the dependency tree.

For conda's search it would be helpful to remove the automatic wildcard, but the same wildcard would resolve quickly through the repodata_shards.msgpack.zst index.

% conda search cond            
Loading channels: done
No match found for: cond. Search: *cond*

For whoneeds I was a little surprised that it queries the entire repo; often I want to restrict this kind of query to packages installed locally in a particular environment.

[dholth]

I don't see a companion issue for "conda search" (not conda repoquery search) "should use sharded repodata"