fix: Validate chart metric column for non-count aggregations

[sentry]

This PR addresses the ValueError: Column is required for avg aggregation issue (DALGO-BACKEND-2GW) by implementing early input validation for chart metrics.

Previously, the API allowed requests with non-count aggregations (e.g., avg) to be submitted with an empty string for the column field. This would pass initial Pydantic validation but lead to a ValueError at runtime in ddpui/core/charts/charts_service.py. This runtime error was then caught and logged as an error in Sentry, generating unnecessary alerts for what is fundamentally a client-side input validation problem.

Changes include:

• Added a @model_validator(mode='after') to the ChartMetric schema in ddpui/schemas/chart_schema.py. This validator now:
  • Normalizes any empty string column values to None.
  • Raises a ValueError if a non-count aggregation is specified without a valid column (i.e., column is None).
• Downgraded the logger.error to logger.warning in the ValueError catch blocks within ddpui/api/charts_api.py for chart data generation endpoints. This ensures that validation-related errors are logged appropriately without triggering high-severity Sentry alerts.

This ensures that invalid chart metric configurations are rejected with a clear 422 HTTP response at the API boundary, improving API robustness and reducing Sentry noise. Existing runtime checks in charts_service.py remain as a defense-in-depth.

Fixes DALGO-BACKEND-2GW