SOLR-17760: solving bug in LTR dense/sparse format

[aruggero]

https://issues.apache.org/jira/browse/SOLR-17760

Description

A bug in the LTR dense/sparse format has been fixed with this PR.
Dense/sparse format was not working for some features (e.g. FieldValueFeature) since default values were returned in the sparse format.
A check if the scored feature has the default value has been added. If it is, the value is excluded from logging in the sparse format.

Solution

The if (featuresInfo[featureId].getValue() != scFW.getDefaultValue()) condition has been added in both the org.apache.solr.ltr.LTRScoringQuery.ModelWeight.ModelScorer.SparseModelScorer#score method of the SparseModelScorer and in the org.apache.solr.ltr.LTRScoringQuery.ModelWeight.ModelScorer.DenseModelScorer#score method of the DenseModelScorer in the LTRScoringQuery class.

Also, the isUsed variable of the org.apache.solr.ltr.LTRScoringQuery.FeatureInfo class has been renamed to isDefaultValue for more clarity.

Finally, the documentation has been updated, integrating the defaultValue parameter, which is a parameter supported by all the LTR features but was not present and explained in the reference guide.

Tests

TestSelectiveWeightCreation and TestFieldValueFeature have been adapted to reflect the fixed behaviour.

Checklist

Please review the following and check all that apply:

• I have reviewed the guidelines for How to Contribute and my code conforms to the standards described there to the best of my ability.
• I have created a Jira issue and added the issue ID to my pull request title.
• I have given Solr maintainers access to contribute to my PR branch. (optional but recommended, not available for branches on forks living under an organisation)
• I have developed this patch against the main branch.
• I have run ./gradlew check.
• I have added tests for my changes.
• I have added documentation for the Reference Guide

[alessandrobenedetti]

this will likely go into 9.9, please change it (we can also do it just before the merge).
If I merge it I'll also add 'via Alessandro Benedetti' that indicates the committer that reviewed and merged

[alessandrobenedetti]

Some minor changes should be done, reverting tests unless necessary and fixing the CHANGES.TXT

[alessandrobenedetti]

Are all these tests just refactoring? It's hard to review, and if there's no actual test change, I would strongly encourage leaving them as they were.
When bug-fixing, the less we touch the tests, the better, so it's clear nothing has been broken by the bugfix,

[aruggero]

This is needed since, depending on the sparse/dense format, we could have default values to be returned.
Before the bug fix, zero values were returned even with the sparse format.
With these changes, we are aligning the expected output of the tests to the randomly selected feature format.