merge eats CPU much when there are many deleteByQuery [LUCENE-7049]

## When
adding very many delete><query>

## Then
we got CPU spike in merge thread that blocks indexing process

## Considerations
Despite adding too many <delete><query> is odd itself, I suppose [the code can more efficient](https://issues.apache.org/jira/browse/LUCENE-5666?focusedCommentId=15129040&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15129040). See sampling snapshots attached.




---
Migrated from [LUCENE-7049](https://issues.apache.org/jira/browse/LUCENE-7049) by Mikhail Khludnev (@mkhludnev), updated Aug 10 2016
Attachments: [Selection_114.png](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-7049/Selection_114.png), [Selection_115.png](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-7049/Selection_115.png), [Selection_116.png](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-7049/Selection_116.png)
Linked issues:
- #6728
- [SOLR-9398](https://issues.apache.org/jira/browse/SOLR-9398)

Start with the reported behavior: add many delete-by-query operations and observe CPU usage in the merge thread while indexing. Review the linked LUCENE-5666 discussion, #6728, and SOLR-9398 for context and possible scope. Done would require a demonstrated reduction in merge CPU contention, supported by a regression test or reproducible measurements.

Written by the indexing model from the issue text.