[Feat] Allowing users to set several quality evaluators

[yiying-zhang]

Users can specify a set of quality metrics, with soft or hard requirements and priorities.

For example, a workflow's quality can be measured by

1. (highest priority) answer accuracy
2. (medium priority) answer tone
3. (hard metric) no hate speech

This requirements include adding interface for users to specify multiple quality metrics and integration to the optimizer for multi-objective, priority-based consideration