Executes incremental reduce in the search thread pool

[jimczi]

This change forks the execution of partial reduces in the coordinating node to the search thread pool.
It also ensures that partial reduces are executed sequentially and asynchronously in order to limit the
memory and cpu that a single search request can use but also to avoid blocking a network thread.
If a partial reduce fails with an exception, the search request is cancelled and the reporting of the error is
delayed to the start of the fetch phase (when the final reduce is performed). This ensures that we cleanup the
in-flight search requests before returning an error to the user.

Closes #53411
Relates #51857

[elasticmachine]

Pinging @elastic/es-search (:Search/Search)

[javanna]

I left a couple of minor comments/questions. I think I will have to look at it another 3/4 times before I can properly review it :)

[nik9000]

I left a few things, mostly small. I think this is the sort of thing that we need, but I also feel like I don't know JVM threading stuff well enough to be super clear. I feel like we're being fairly paranoid here with the threading stuff. Which is probably fine because this won't trigger on small searches. But I don't know enough to be sure that my instinct is right about it either.

[jimczi]

Thanks for looking @nik9000 @javanna . I pushed another commit to address your comments and to simplify the change a bit. Would you mind taking another look ?

[nik9000]

I had another look this morning and I think I mostly understand it. I left two small things.