Summary

This PR introduces a batchSize parameter to RedisItemReader and its builder to solve the classic N+1 issue faced during batch reads from Redis.
Currently, RedisItemReader executes individual GET commands per Redis key resulting in a significant number of network round-trips and performance bottlenecks.

By enabling batch reading with Redis MGET when batchSize > 1, this enhancement drastically reduces network calls and improves throughput while maintaining full backward compatibility when batchSize is not set (default = 1).

Key Changes

• Added batchSize property with default value 1 in RedisItemReader and RedisItemReaderBuilder.
• When batchSize > 1, the reader fetches multiple keys in batch using Redis multiGet() and internally buffers results.
• Existing behavior preserved for batchSize = 1, ensuring no breaking changes.
• Added thorough JavaDoc explaining usage, limitations, and batch size considerations.
• Updated unit tests to cover both single and batch modes, verifying correct data retrieval and network call optimization.

Benefits

• Significant performance improvements through reduced network round-trips during batch processing.
• Configurable batching allowing tuning based on memory and performance requirements.
• Maintains Spring Batch's principle of backward compatibility and progressive enhancement.
• Handles Redis SCAN cursor nature without breaking existing semantics.

Technical Considerations

• Batch size validated to be greater than 0; recommending values between 1 and 1000 to prevent excessive memory use and large network calls.
• The internal queue buffers values for efficient retrieval during batch reads.
• Users should be aware of Redis SCAN characteristics such as key duplication and lack of ordering; idempotent batch jobs recommended.
• Restartability remains unsupported due to unordered nature of Redis SCAN results.

Related Issue

Fixes #4941

"Terminated by KILL-9 SQUAD 💀"