[C++][Acero] Support for multi threaded input: ScalarAggregateNode , GroupByNode

[X-Lemon-X]

Describe the enhancement requested

• Support for multi threading inputs (batches incoming not in order if they were ordered) with nodes like ScalarAggregateNode , GroupByNode and SortedMergeNode, previously resolved by disabling multi threading when using them.
• Propagating ordering information on output of nodes ScalarAggregateNode and GroupByNode when input is ordered.

I also created a PR with all the changes for this.

Component(s)

C++
Acero