Optimized Chinese search when NLQ is enabled.

[MochiXu]

Currently, with NLQ enabled, Tantivy treats "伊文斯隐瞒秘密" as a PhraseQuery, which is its default behavior. We need to refine this for Chinese. Here's a possible modification method:

query = "(Who is Obama?) OR TS\"伊文斯隐瞒秘密\""

By using "TS" to mark a subquery, the NLQ parser will generate a TermSetQuery for TS"伊文斯隐瞒秘密".