Optimize order of file opening to match query

### Is your feature request related to a problem or challenge?

Since we now have dynamic predicate pushdown from TopK operators and file statistics by default it makes sense that we use those statistics if available to optimize the order we scan files in.

Context: https://github.com/apache/datafusion/issues/16402#issuecomment-3005250264

Start with the linked context in issue 16402, then trace the existing dynamic predicate pushdown from TopK operators and the file-statistics path. The change is complete when available statistics influence file scan order and the relevant query behavior is covered by tests, though this issue names no specific files or test targets.

Written by the indexing model from the issue text.