feat: Intelligence Pipeline v3 — full implementation

Multi-stage evidence-grounded inference architecture replacing the
monolithic 9B model extraction pipeline. CPU-first specialist services
handle routine extraction while the 9B vLLM model is preserved for
semantic adjudication of ambiguous cases.

Key components:
- Capability-aware inference gateway (OpenAI-compatible + Ollama)
- Endpoint registry with DB migrations and REST API
- Sentence-aware document segmenter (property tests)
- Deterministic financial parsing with offset integrity
- Symbol resolution with ambiguity detection
- Specialist service (GLiNER2, dynamic batching, K8s deployment)
- Company-specific sentiment (FinBERT, calibration)
- Retrieval-based novelty and duplicate detection
- Confidence calibration pipeline
- Deterministic routing engine (property tests)
- 9B adjudication layer with VRAM gating
- Stock-specific impact model (features, labels, baseline, trained)
- Pipeline orchestrator (state machine, queues, leases, feature flags)
- Bounded parallelism (async workers, semaphore, load shedding)
- Observability (tracing, metrics, alerts)
- Compatibility adapter (v3→v2 golden mapping tests)
- Shadow/canary promotion framework
- Active learning and fine-tuning pipeline

Test results: 1,161 tests pass, ruff lint clean.
All 282 spec tasks completed.

services/specialist/models.py

@@ -0,0 +1,113 @@
+"""Request and response models for the specialist inference service."""
+
+from __future__ import annotations
+
+from pydantic import BaseModel, Field
+
+# ---------------------------------------------------------------------------
+# Constants
+# ---------------------------------------------------------------------------
+
+MODEL_VERSION = "gliner2-large-v1.0"
+SCHEMA_VERSION = "specialist-v1"
+
+
+# ---------------------------------------------------------------------------
+# Request models
+# ---------------------------------------------------------------------------
+
+
+class ExtractionRequest(BaseModel):
+    """Batch extraction request accepted by entity, relation, and structured endpoints."""
+
+    texts: list[str] = Field(..., min_length=1, description="Texts to process")
+    schema_labels: list[str] = Field(
+        ..., min_length=1, description="Entity/relation/event labels to extract"
+    )
+    batch_id: str | None = Field(
+        default=None, description="Optional caller-provided batch identifier"
+    )
+
+
+class ClassificationRequest(BaseModel):
+    """Batch classification request accepted by the classify endpoint."""
+
+    texts: list[str] = Field(..., min_length=1, description="Texts to classify")
+    schema_labels: list[str] = Field(
+        ..., min_length=1, description="Classification labels"
+    )
+    batch_id: str | None = Field(
+        default=None, description="Optional caller-provided batch identifier"
+    )
+
+
+# ---------------------------------------------------------------------------
+# Result models
+# ---------------------------------------------------------------------------
+
+
+class EntityResult(BaseModel):
+    """A single extracted entity span."""
+
+    text: str
+    entity_type: str
+    start_char: int
+    end_char: int
+    score: float = Field(..., ge=0.0, le=1.0)
+    model_version: str = MODEL_VERSION
+    schema_version: str = SCHEMA_VERSION
+
+
+class RelationResult(BaseModel):
+    """A single extracted relation."""
+
+    subject: str
+    subject_type: str
+    subject_start: int
+    subject_end: int
+    relation: str
+    object: str
+    object_type: str
+    object_start: int
+    object_end: int
+    score: float = Field(..., ge=0.0, le=1.0)
+    model_version: str = MODEL_VERSION
+    schema_version: str = SCHEMA_VERSION
+
+
+class ClassificationResult(BaseModel):
+    """A single classification result."""
+
+    text: str
+    label: str
+    score: float = Field(..., ge=0.0, le=1.0)
+    model_version: str = MODEL_VERSION
+    schema_version: str = SCHEMA_VERSION
+
+
+class StructuredResult(BaseModel):
+    """A single structured extraction result with key-value facts."""
+
+    text: str
+    field: str
+    value: str
+    start_char: int
+    end_char: int
+    score: float = Field(..., ge=0.0, le=1.0)
+    model_version: str = MODEL_VERSION
+    schema_version: str = SCHEMA_VERSION
+
+
+# ---------------------------------------------------------------------------
+# Batch response
+# ---------------------------------------------------------------------------
+
+
+class BatchResponse(BaseModel):
+    """Unified batch response wrapping results from any endpoint."""
+
+    results: list[list[EntityResult]] | list[list[RelationResult]] | list[list[ClassificationResult]] | list[list[StructuredResult]]
+    model_version: str = MODEL_VERSION
+    schema_version: str = SCHEMA_VERSION
+    processing_time_ms: float
+    batch_id: str | None = None