doc: supplement dirty read prevention for query scenarios

[Oxidaner]

What is this PR for?

This PR supplements the missing explanation of how to prevent dirty reads in Seata query scenarios in the current documentation (e.g., the section related to updateAll() and queryA()), as the existing content appeared incomplete.

I added two common scenarios for dirty read prevention (for queryA()), along with corresponding solutions, execution logic, and code examples, to make the documentation more comprehensive and actionable for users.

What changes were made?

1. Added Scenario 1: How to ensure queryA() accurately reads the final committed data of global transactions (using @GlobalLock + Read Committed isolation level + select ... for share), with a code snippet.
2. Added Scenario 2: How to prevent dirty reads for regular queries (using Read Committed isolation level), explaining the execution logic.

Does this PR introduce any breaking changes?

• Yes
• ✅ No

Checklist

• ✅ The added content aligns with Seata's actual behavior (tested with AT mode + @GlobalLock).
• ✅ The explanation is consistent with existing documentation terminology (e.g., "global lock", "Read Committed").
• ✅ Both Chinese and English versions (if applicable) are supplemented