Make dynamic range facets value collection and sorting faster

### Description

`DynamicRangeUtil` [collects](https://github.com/apache/lucene/blob/02b37670e481679363e8ae5c245e5ed3cb0f10b6/lucene/facet/src/java/org/apache/lucene/facet/range/DynamicRangeUtil.java#L145) values from each segment and then [sorts](https://github.com/apache/lucene/blob/02b37670e481679363e8ae5c245e5ed3cb0f10b6/lucene/facet/src/java/org/apache/lucene/facet/range/DynamicRangeUtil.java#L213) all the values in the main thread. I wonder if we could get a speed-up from collecting values in each segment, sorting them (maybe doing insertion sort), and then merging the sorted values from all segments, effectively moving more of the work to the executor and doing less in the main thread.

Start in lucene/facet/src/java/org/apache/lucene/facet/range/DynamicRangeUtil.java at the collection point around line 145 and sorting point around line 213. Trace how segment work is submitted to the executor, then evaluate segment-level sorting and merging of the results. Done means value collection and sorting are moved out of the main-thread bottleneck without changing dynamic range facet behavior.

Written by the indexing model from the issue text.