feat: bake sampling_rate into Judge at construction; simplify Evaluator to List[Judge]

[jsonbailey]

Summary

• Judge.__init__ gains sample_rate: float = 1.0; stored as self.sample_rate
• Judge.evaluate(sampling_rate=None) and evaluate_messages(sampling_ratio=None) fall back to self.sample_rate when the arg is omitted; an explicit 0.0 still overrides correctly
• Evaluator constructor simplified to List[Judge] only — JudgeConfiguration removed from the constructor and evaluate path
• Evaluator.noop() returns cls([])
• _initialize_judges deleted; _build_evaluator inlined via a private _create_judge_instance helper shared with create_judge so the public create_judge surface is unchanged and no usage-tracking event is emitted during internal evaluator construction
• 4 new tests covering sample_rate default, explicit value, instance-rate fallback, and per-call override

Test plan

• 146 server-ai tests pass
• 81 langchain tests pass, 40 openai tests pass (Evaluator.noop() callers unaffected)
• mypy, isort, pycodestyle clean

🤖 Generated with Claude Code

---

Note

Medium Risk
Changes how judge sampling rates are applied and refactors Evaluator/judge construction, which can alter when evaluations run and could affect tracking or performance if misconfigured. Scope is limited to judge/evaluator plumbing and is covered by added unit tests.

Overview
Bakes per-judge sampling into Judge instances by adding a sample_rate constructor arg and making Judge.evaluate()/evaluate_messages() default to the instance rate when no per-call rate is provided.

Simplifies Evaluator to hold a List[Judge] (removing JudgeConfiguration from the execution path) and updates LDAIClient._build_evaluator() to materialize judges inline via a new _create_judge_instance() helper that avoids emitting the public create-judge usage event during internal evaluator construction.

Adds tests covering sample_rate defaulting, constructor override, instance-rate fallback, and per-call override behavior.

^{Reviewed by Cursor Bugbot for commit 13b38fd. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 2d80afe. Configure here.}