[Evals API][6/n] meta-reference llm as judge, registration for ScoringFnDefs (#330)

* wip scoring refactor

* llm as judge, move folders

* test full generation + eval

* extract score regex to llm context

* remove prints, cleanup braintrust in this branch

* change json -> class

* remove initialize

* address nits

* check identifier prefix

* udpate MANIFEST

llama_stack/apis/scoring_functions/scoring_functions.py

@@ -26,6 +26,10 @@ class Parameter(BaseModel):
 class LLMAsJudgeContext(BaseModel):
     judge_model: str
     prompt_template: Optional[str] = None
+    judge_score_regex: Optional[List[str]] = Field(
+        description="Regex to extract the score from the judge response",
+        default=None,
+    )
 
 
 @json_schema_type