feat: pydantic schemas for all service message types

- shared/schemas/trading.py: OrderRequest, OrderResult, PositionInfo,
  AccountInfo, TradeSignal, TradeExecution, MarketSnapshot, SentimentContext
- shared/schemas/news.py: RawArticle, ScoredArticle
- shared/schemas/learning.py: TradeOutcomeSchema, WeightAdjustment
- shared/schemas/auth.py: RegisterRequest, LoginRequest, TokenResponse
- 49 schema tests covering validation constraints, serialization round-trips,
  required fields, and range checks

shared/schemas/learning.py

@@ -0,0 +1,32 @@
+"""Learning domain Pydantic schemas."""
+
+from datetime import datetime
+from uuid import UUID
+
+from pydantic import BaseModel, Field
+
+
+class TradeOutcomeSchema(BaseModel):
+    """Represents the evaluated outcome of a closed trade."""
+
+    trade_id: UUID
+    hold_duration_seconds: float = Field(ge=0)
+    realized_pnl: float
+    roi_pct: float
+    was_profitable: bool
+
+    model_config = {"from_attributes": True}
+
+
+class WeightAdjustment(BaseModel):
+    """Represents a strategy weight change made by the learning engine."""
+
+    strategy_id: UUID
+    strategy_name: str
+    old_weight: float
+    new_weight: float
+    reason: str
+    reward_signal: float
+    timestamp: datetime
+
+    model_config = {"from_attributes": True}