feat: implement validation to forbid dialog rails with reasoning traces

[Pouyanpi]

Description

This PR implements a validator to ensure that reasoning traces cannot be enabled when dialog rails are present. The validator checks for both explicit dialog rails configuration and implicit dialog rails enabled through user messages, bot messages, and flows.

Impact

This change ensures that:

1. Users cannot enable reasoning traces when dialog rails are present
2. The validation covers all ways dialog rails can be enabled
3. Clear error messages guide users to resolve configuration conflicts

Changes

1. Added check_reasoning_traces_with_dialog_rails validator to RailsConfig class that:

   • Checks if any model has reasoning traces enabled
   • Verifies presence of dialog rails through:
     • Explicit configuration in rails.dialog
     • Implicit configuration through user_messages
     • Implicit configuration through bot_messages
     • Implicit configuration through flows
   • Raises a ValueError if reasoning traces are enabled while dialog rails are present

2. Added test cases in test_config_validation.py to verify:

   • Error is raised when both reasoning traces and dialog rails are enabled
   • No error when only reasoning traces are present
   • No error when only dialog rails are present
   • Error is raised when reasoning traces are enabled with implicit dialog rails

Testing

Added comprehensive test coverage:

Technical Details

The validator is implemented as a root validator in the RailsConfig class that (must migrated as part of #967:

• Takes pre-validated values as input
• Checks model configurations for reasoning traces
• Verifies dialog rails presence through multiple sources
• Returns the values if validation passes
• Raises an error if validation fails

Breaking Changes

None. This is a new validation that prevents invalid configurations while it maintains backward compatibility

TODO:

• check for reasoning traces per dialog task