refactor: reorganize source files into core/, evaluation/, runners/, scenarios/ directories

Moved loose root-level Python modules into logical subdirectories:
- core/ — models, client, episode_store
- evaluation/ — grading, agent_brutal_audit
- runners/ — baseline_runner, inference
- scenarios/ — simulation, case_generator, iso_adapter

core/__init__.py

@@ -0,0 +1 @@
+"""Core data models, client, and storage for ChargebackOps."""

evaluation/__init__.py

@@ -0,0 +1 @@
+"""Grading and audit modules for ChargebackOps."""

agent_brutal_audit.py → evaluation/agent_brutal_audit.py

@@ -21,11 +21,11 @@ import statistics
 from pathlib import Path
 from typing import Any
 
-from baseline_runner import _heuristic_pick, _obvious_next_action, candidate_actions
-from grading import grade_episode
-from models import ChargebackOpsAction
+from runners.baseline_runner import _heuristic_pick, _obvious_next_action, candidate_actions
+from evaluation.grading import grade_episode
+from core.models import ChargebackOpsAction
 from server.chargeback_ops_environment import ChargebackOpsEnvironment
-from simulation import get_task
+from scenarios.simulation import get_task
 
 DATA_DIR = Path("data")
 

grading.py → evaluation/grading.py

@@ -3,11 +3,11 @@
 from __future__ import annotations
 
 try:
-    from .models import CaseScoreBreakdown, GraderReport
-    from .simulation import CaseProgress, InternalCase, TaskScenario
+    from ..core.models import CaseScoreBreakdown, GraderReport
+    from ..scenarios.simulation import CaseProgress, InternalCase, TaskScenario
 except ImportError:  # pragma: no cover
-    from models import CaseScoreBreakdown, GraderReport
-    from simulation import CaseProgress, InternalCase, TaskScenario
+    from core.models import CaseScoreBreakdown, GraderReport
+    from scenarios.simulation import CaseProgress, InternalCase, TaskScenario
 
 
 def _ratio(numerator: int, denominator: int) -> float:

runners/__init__.py

@@ -0,0 +1 @@
+"""Baseline and inference runners for ChargebackOps."""

baseline_runner.py → runners/baseline_runner.py

@@ -12,15 +12,15 @@ from openai import OpenAI
 from pydantic import BaseModel, Field
 
 try:
-    from .grading import grade_episode
-    from .models import BaselineRunResult, BaselineTaskResult, ChargebackOpsAction
-    from .server.chargeback_ops_environment import ChargebackOpsEnvironment
-    from .simulation import list_tasks
+    from ..evaluation.grading import grade_episode
+    from ..core.models import BaselineRunResult, BaselineTaskResult, ChargebackOpsAction
+    from ..server.chargeback_ops_environment import ChargebackOpsEnvironment
+    from ..scenarios.simulation import list_tasks
 except ImportError:  # pragma: no cover
-    from grading import grade_episode
-    from models import BaselineRunResult, BaselineTaskResult, ChargebackOpsAction
+    from evaluation.grading import grade_episode
+    from core.models import BaselineRunResult, BaselineTaskResult, ChargebackOpsAction
     from server.chargeback_ops_environment import ChargebackOpsEnvironment
-    from simulation import list_tasks
+    from scenarios.simulation import list_tasks
 
 try:  # pragma: no cover
     from dotenv import load_dotenv

inference.py → runners/inference.py

@@ -25,12 +25,12 @@ try:
         _strict_llm_mode,
         candidate_actions,
     )
-    from .grading import grade_episode
-    from .models import BaselineRunResult, BaselineTaskResult
-    from .server.chargeback_ops_environment import ChargebackOpsEnvironment
-    from .simulation import list_tasks
+    from ..evaluation.grading import grade_episode
+    from ..core.models import BaselineRunResult, BaselineTaskResult
+    from ..server.chargeback_ops_environment import ChargebackOpsEnvironment
+    from ..scenarios.simulation import list_tasks
 except ImportError:  # pragma: no cover
-    from baseline_runner import (
+    from runners.baseline_runner import (
         MAX_PROVIDER_RESPONSE_TOKENS,
         _chat_completion_with_retry,
         _heuristic_pick,
@@ -40,10 +40,10 @@ except ImportError:  # pragma: no cover
         _strict_llm_mode,
         candidate_actions,
     )
-    from grading import grade_episode
-    from models import BaselineRunResult, BaselineTaskResult
+    from evaluation.grading import grade_episode
+    from core.models import BaselineRunResult, BaselineTaskResult
     from server.chargeback_ops_environment import ChargebackOpsEnvironment
-    from simulation import list_tasks
+    from scenarios.simulation import list_tasks
 
 if load_dotenv is not None:  # pragma: no cover
     load_dotenv()

scenarios/__init__.py

@@ -0,0 +1 @@
+"""Task scenarios, case generation, and ISO adapters for ChargebackOps."""