Follow up improvement for reverse parquet when row group size is huge, so it may cause OOM.

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

Details:
https://github.com/apache/datafusion/pull/18817#discussion_r2576073230

### Describe the solution you'd like

_No response_

### Describe alternatives you've considered

_No response_

### Additional context

_No response_

Start by reading the discussion linked from the issue, including the referenced DataFusion pull request, to understand the reverse-Parquet improvement and the huge-row-group OOM scenario. Locate the implementation and tests identified there, reproduce or characterize the memory problem, and confirm that the follow-up handles large row groups without OOM.

Written by the indexing model from the issue text.