LLM critic validator enhancement

[rkritika1508]

Describe the bug
On calling LLM critic validator, sometimes we get responses like -

{
  "success": false,
  "data": {
    "response_id": "11c2d6b7-6f2e-4d09-a78a-23b3d40aed6b",
    "rephrase_needed": false
  },
  "error": "The response failed the following metrics: ['query']."
}

And sometimes like this -

{
  "success": false,
  "data": {
    "response_id": "4985fe71-39fb-42d4-977c-d2d27a16f2d6",
    "rephrase_needed": false
  },
  "error": "The response is missing or has invalid evaluations for the following metrics: ['query']."
}

Additional context
We are going to investigate why that happens and override the errors for the validator. The end goal is to provide proper exception to the end users.