Add optional JSON-array output for the Agent Flow Iteration node

[Dexterity104]

Summary

Adds an Output Structure option on the Agent Flow Iteration node so results can stay in a fixed order and length that matches the input array, instead of only a single joined text string.

It's useful for flows that treat the iteration as a map over sections or rows and need a stable list to feed into another Iteration or any step that expects parseable JSON.

Closes #6091

Behavior

• Aggregated Text (default): unchanged. output.content is all iteration outputs joined with newlines.
• JSON Array: output.content is a JSON string of a string array (one entry per input index). If a sub-run has no text, that index is still present as an empty string so positions stay aligned with the input.

output.iterationResults remains the underlying string[] in both modes. content stays a string everywhere so the usual flow completion and variable paths keep working.

[gemini-code-assist]

Code Review

This pull request introduces a new Row Text Splitter node for line-based text processing and adds an Output Structure option to the Iteration node, allowing users to choose between aggregated text and JSON array formats. The server-side execution logic was updated to support these structures and ensure index alignment for JSON array outputs. I have no feedback to provide.