Optimize GroupBy.Sum and GroupBy.Mean for large DataFrames

[dwalterskoetter]

I've analyzed the GroupBy.cs implementation and identified opportunities for performance improvements, including:

• Implement single-pass iteration over the group rows.
• Introduce typed accumulators for double and long to avoid boxing and improve numeric performance.
• Pre-allocate result columns (PrimitiveDataFrameColumn) to reduce repeated resizing.
• Optimize delegates used during iteration for better performance.

I noticed issue #6824 tracks DataFrame performance improvements. Would performance enhancements for GroupBy operations be welcome as part of this effort?

I can provide benchmarks and a PR if there's interest.