feat(memory): implement memory reflection service (#99)

Add reflection layer for memory system with pattern detection, success/failure
factor analysis, anomaly detection, and insights generation. Enables agents to
learn from past experiences and identify optimization opportunities.

Key components:
- Pattern detection: recurring success/failure, action sequences, temporal, efficiency
- Factor analysis: action, context, timing, resource, preceding state factors
- Anomaly detection: unusual duration, token usage, failure rates, action patterns
- Insight generation: optimization, warning, learning, recommendation, trend insights

Also fixes pre-existing timezone issues in test_types.py (datetime.now() -> datetime.now(UTC)).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

backend/app/services/memory/reflection/__init__.py

@@ -0,0 +1,38 @@
+# app/services/memory/reflection/__init__.py
+"""
+Memory Reflection Layer.
+
+Analyzes patterns in agent experiences to generate actionable insights.
+"""
+
+from .service import (
+    MemoryReflection,
+    ReflectionConfig,
+    get_memory_reflection,
+)
+from .types import (
+    Anomaly,
+    AnomalyType,
+    Factor,
+    FactorType,
+    Insight,
+    InsightType,
+    Pattern,
+    PatternType,
+    TimeRange,
+)
+
+__all__ = [
+    "Anomaly",
+    "AnomalyType",
+    "Factor",
+    "FactorType",
+    "Insight",
+    "InsightType",
+    "MemoryReflection",
+    "Pattern",
+    "PatternType",
+    "ReflectionConfig",
+    "TimeRange",
+    "get_memory_reflection",
+]