Improve performance of SortPreservingMergeExec

[NGA-TRAN]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
While doing experience in IOx, we have discovered that SortPreservingMergeExec is expensive and very CPU bound. The issues are fully described here

Describe the solution you'd like
The SortPreservingMergeExec is known the be sub-optimal when it was implemented and at least 3 improvements have been identified inside the code. This ticket is to improve those and further if any.

[alamb]

I believe @e-dard has plans to look into this ticket