refactor: show warnings for AI responses

[kevinkim-ogp]

TL;DR

Added ai_response type to identify AI-generated variables and display appropriate warnings to users.

This change helps users identify AI-generated content and reminds them to verify the outputs before using them in critical workflows. By clearly marking AI responses and displaying appropriate warnings, we reduce the risk of users blindly trusting potentially incorrect AI-generated content.

What changed?

• Added a new ai_response type to the TDataOutMetadatumType enum
• Created getDataOutMetadata for the Pair action + updated AISAY's getDataOutMetadata
• Added UI elements to warn users about AI-generated content:
  • Warning icon and message in the FlowStepTestController for AI steps
  • Special tag and tooltip for AI response variables in the VariablesList

How to test?

Create a flow with an AI step (either Pair or AISAY)

• Run a test on the step and observe the warning message: "AI can make mistakes. Check your responses across multiple test cases."
• Check that variables from AI steps are tagged as "AI response" with the tooltip: "This variable is AI-generated. Check for errors before using it."

Screenshots

[kevinkim-ogp]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• chore: expand ai test result #1325
• refactor: show warnings for AI responses #1316 👈 (View in Graphite)
• [PAIR-2]: Pair action structured response #1315
• [PAIR-1]: Pair action single response #1314
• [AI-ACTION-2]: Add tiptap-markdown to get markdown #1313 : 1 other dependent PR (#1324 )
• [AI-ACTION-1]: Refactor RTE to allow custom menu options #1312
• [AI-2]: Bump typescript and eslint deps #1301 : 1 other dependent PR (#1302 )
• [AI-1]: Install deps and bump zod #1300
• trunk/ai

This stack of pull requests is managed by Graphite. Learn more about stacking.