Request: Option to skip bigrams indexing across sentence boundary

Nobody has claimed this yet.

**Is your feature request related to a problem? Please describe.**

On a standard index, can use `phrase_boundary` config to stop normal phrase queries from matching across sentence boundaries.

But when using `bigram_index`, the indexed bigrams (in the dictionary) include bigrams across sentence boundaries. So now phrase searches cross boundaries.

**Describe the solution you'd like**

For a way to have bigram indexing honour `phrase_boundary` setting, and hence NOT index bigrams that are seperated by chars in `phrase_boundary`.

bigram_index = non_boundary

I guess, ideal would be to use with `first_freq` or `both_freq` but personally only want it with `all`. So `non_boundary` would mean 'all except phrase_boundary seperated pairs'

I suppose it could also be based in `index_sp`, but frankly less ideal, as it not as configurable. (eg can't make comma a phrase boundary!) ... and its based on the html stipper, which may not be needed or wanted.

**Describe alternatives you've considered**

Could perhaps use regex_filter (or similar) to actually inject a (fake) word in the middle, to upset bigram indexing?

regexp_filter = \b;\s => \t_SEP\t

or somehting (untested!)

**Additional context**

https://forum.manticoresearch.com/t/preventing-bigrams-across-sentence-boundary/799/2

No source file or test is named. Start by tracing the bigram_index configuration and how phrase_boundary is applied during indexing; compare the existing all, first_freq, and both_freq modes. Done means a non_boundary option prevents indexing bigrams separated by configured phrase_boundary characters while preserving current modes.

Written by the indexing model from the issue text.