Summary

Closes #3946

The existing fault-tolerant step documentation (configuring-skip and retry-logic) did not explain chunk scanning, which is the mechanism Spring Batch uses to identify which individual item caused a write failure. This caused confusion: users who set NeverRetryPolicy expected no re-processing to occur, but were surprised when Spring Batch still re-invoked their processor multiple times per chunk.

Changes

Added a new chunk-scanning.adoc page (linked in nav after retry-logic.adoc) covering:

1. How chunk scanning works: on a write exception → rollback → re-process each item individually → skip or fail per-item based on skip policy
2. Key insight: chunk scanning is triggered regardless of retry policy (even with NeverRetryPolicy)
3. Impact on ItemProcessor: the processor is re-invoked for each scanned item; use processorNonTransactional() to avoid this (with Java and XML examples)
4. Relationship to skip and retry: these three mechanisms are orthogonal — chunk scanning enables per-item skip; retry operates within the scan phase

Also added:

• Nav entry in nav.adoc after retry-logic.adoc
• A NOTE cross-reference at the bottom of configuring-skip.adoc pointing to the new page

Test plan

• Docs build with Antora renders new page and nav entry correctly
• Cross-reference link in configuring-skip.adoc resolves
• Java/XML examples are syntactically correct