Extend cases for #11643 - optimization for string terms aggregation

[sandeshkr419]

#11643

W.r.t to the above PR, when optimizing the string terms aggregation for single terms, we omitted some cases:

• Deleted documents in a segment
• Doc frequency exists in any document of a segment

It should be possible that with the optimization, deleted documents in a segment can be handled in a segment gracefully by computing the aggregations using the same optimization and then traversing only through deleted documents.

Need to figure out:

• What should be those ideal thresholds for the deleted documents compared to the total number of documents in a segment. This will require some data collection with experimenting different percentages of deleted documents.

Note, that we are mentioning this per segment basis as the optimization (if or not) is also relevant at a segment level.

[peternied]

[Triage - attendees 1 2 3 4 5 6]
@sandeshkr419 Thanks for creating this issue, look forward to seeing a pull request to resolve.

From @peternied please use the template for issue creation