[Refactor] Consolidate DataFrame formatting logic in FlintJob

[manaslohani]

Currently, DataFrame formatting in FlintJob is scattered across multiple locations:

• SingleStatementExecutionManagerImpl handles DataFrame formatting for streaming queries
• QueryResultWriter.processDataFrame() handles for successful interactive queries
• constructErrorDF handles all failed queries

Consolidating all DataFrame formatting through a single unified approach using QueryResultWriter.processDataFrame method. This will centralize all DataFrame formatting logic in one place.

Notes

The following considerations need to be addressed during implementation:

• Pass jobType to processDataFrame for different cleanup logic between streaming and interactive queries
• ** Refactor logic for getting requestIndex/ Ensure requestIndex is available in session context** for streaming queries to maintain current functionality
• Handle failed/timeout states generically - either create a new method or extend the existing processDataFrame method to handle both success and failure cases

[manaslohani]

Could also revisit the metadata dataframe schema to be written here for the result Index.