Remove duplicated batchSize factor from target load; add a test case

[cloudhunter89]

Addresses a bug identified in a slack thread (https://confluentcommunity.slack.com/archives/C051K2TJLPQ/p1699018828805729) which causes the inflight queue to scale by batchSize * batchSize * maxConcurrency * dynamicLoadFactor rather than batchSize * maxConcurrency * dynamicLoadFactor

Checklist

• Documentation (if applicable)
• Changelog

[cla-assistant]

All committers have signed the CLA.

[rkolesnev]

LGTM - just one nitpick

[what-the-diff]

PR Summary

• Updated Target Loading Computation for Inflight Records
  A tweak has been made to the relevant process in the CHANGELOG.adoc so that it computes the target loading for inflight records in a more efficient manner.

• Altered getQueueTargetLoaded() Method
  The method getQueueTargetLoaded() in AbstractParallelEoSStreamProcessor.java has been changed, eliminating the multiplication by batch size, which results in more accurate performance.

• New Test File Added
  A new test file AbstractParallelEoSStreamProcessorConfigurationTest.java was introduced to improve the reliability of the given software by enabling us to inspect its behavior under various conditions.

• New File for Testing Internal and Configuration-based Interfaces
  A new file named TestParallelEoSStreamProcessor.java has been added. This provides us with methods for testing the internal functionality and configuration-based adjusting aspects of the AbstractParallelEoSStreamProcessor, ensuring the system works smoothly and correctly.