Adding float64 support, document level boosting, and facet collector

[AliFlux]

This PR adds a couple of new features that are present in Tantivy core repo, but not exposed in tantivy-py:

Conjunction by default parameter

By default, tantivy parses queries using OR operator, instead of AND operator. If we want to modify this behavior, we can now set this value when parsing query:

parse_query(text, fields, conjunction_by_default=True)

Floating point support

A new add_float_field function is available so that we can add f64 fields.

Document level boosting #51

We can now give priority to certain documents using the new weight_by_field parameter:

searcher.search(query, limit, weight_by_field='popularity')

TopDocs tweak_score is used, and the callback is abstracted away from python code for performance reasons.

Facet collector

We can now get counts of facets available by specifying the count_facets_by_field parameter:

data = searcher.search(query, limit, count_facets_by_field='genre')

print(data.facet_counts)
# prints a dictionary where keys are facets, and values are counts

[cjrh]

I'm excited for f64 support to get in. I'm not a maintainer of tantivy-py, but I left several comments hoping that it helps to move forward on the PR.

Perhaps f64 support would get merged faster if you made a separate PR for just the f64 support?

[cjrh]

I was going to comment about the unwrap but I see that the LHS schema::Facet::from is just implemented using from_text with an unwrap anyway: https://docs.rs/tantivy/0.18.0/src/tantivy/schema/facet.rs.html#181-183

I'm curious what was the reason for this change?

[AliFlux]

I don't exactly remember this change. Maybe I couldn't get facets to work without it

[AliFlux]

@poljar let me know if there's anything I can do to expedite the merge

[cjrh]

This file seems like it was added in error...?