Optimize IO by using projection

**Is your feature request related to a problem or challenge? Please describe what you are trying to do.**
After the PR https://github.com/apache/arrow-rs/pull/1339 the projection is supported.
While this saves some copies / array creation, this doesn't use the projection yet avoid IO for reading the data (from a file).

**Describe the solution you'd like**
Use the projection to skip reading parts of the file.

**Describe alternatives you've considered**

**Additional context**
Arrow2 supports this.

Start by tracing the projection support introduced in PR 1339 through the file-reading path, then compare how Arrow2 uses projection to avoid reading unneeded data. The work is done when projected columns are skipped during file reads rather than only during array creation; add or update coverage for that behavior where the existing reader tests reside.

Written by the indexing model from the issue text.