This is a follow-up to #4254. The original PR contained the smart-OR keyset predicate implementation, but it was closed after the fork branch no longer carried the diff. This PR resubmits the change from a dedicated branch and adds the missing edge-case coverage discussed in #4250.

This change optimizes keyset continuation predicates by adding an inclusive bound for the leading sort property and grouping the existing disjunction behind it.

For multi-property keysets, the generated predicate can now use a shape such as:

firstname <= ? AND (firstname < ? OR firstname = ? AND emailAddress < ?)

The fallback path keeps the existing predicate shape when the leading keyset value is null, preserving the current null-handling behavior.

Additional integration coverage walks keyset pages over datasets with duplicate leading sort values and compares the resulting sequence against regular sorted queries. The added tests cover forward scrolling, backward scrolling, nulls-last with mixed sort directions, and Querydsl predicate scrolling.

While adding the Querydsl nulls-last coverage, the null keyset equality branch was corrected to use IS NULL.

Closes #4250
Supersedes #4254

• You have read the Spring Data contribution guidelines.
• You use the code formatters provided here and have them applied to your changes. Don’t submit any formatting related changes.
• You submit test cases (unit or integration tests) that back your changes.
• You added yourself as author in the headers of the classes you touched. Amend the date range in the Apache license header if needed. For new types, add the license header.