feat: add owner's manual OCR pipeline (refs #71)

Implement async PDF processing for owner's manuals with maintenance
schedule extraction:

- Add PDF preprocessor with PyMuPDF for text/scanned PDF handling
- Add maintenance pattern matching (mileage, time, fluid specs)
- Add service name mapping to maintenance subtypes
- Add table detection and parsing for schedule tables
- Add manual extractor orchestrating the complete pipeline
- Add POST /extract/manual endpoint for async job submission
- Add Redis job queue support for manual extraction jobs
- Add progress tracking during processing

Processing pipeline:
1. Analyze PDF structure (text layer vs scanned)
2. Find maintenance schedule sections
3. Extract text or OCR scanned pages at 300 DPI
4. Detect and parse maintenance tables
5. Normalize service names and extract intervals
6. Return structured maintenance schedules with confidence scores

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

ocr/app/models/__init__.py

@@ -6,6 +6,10 @@ from .schemas import (
     JobResponse,
     JobStatus,
     JobSubmitRequest,
+    ManualExtractionResponse,
+    ManualJobResponse,
+    ManualMaintenanceSchedule,
+    ManualVehicleInfo,
     OcrResponse,
     ReceiptExtractedField,
     ReceiptExtractionResponse,
@@ -20,6 +24,10 @@ __all__ = [
     "JobResponse",
     "JobStatus",
     "JobSubmitRequest",
+    "ManualExtractionResponse",
+    "ManualJobResponse",
+    "ManualMaintenanceSchedule",
+    "ManualVehicleInfo",
     "OcrResponse",
     "ReceiptExtractedField",
     "ReceiptExtractionResponse",

ocr/app/models/schemas.py

@@ -115,3 +115,57 @@ class ReceiptExtractionResponse(BaseModel):
     error: Optional[str] = None
 
     model_config = {"populate_by_name": True}
+
+
+# Manual extraction models
+
+
+class ManualVehicleInfo(BaseModel):
+    """Vehicle information extracted from manual."""
+
+    make: Optional[str] = None
+    model: Optional[str] = None
+    year: Optional[int] = None
+
+
+class ManualMaintenanceSchedule(BaseModel):
+    """A single maintenance schedule entry."""
+
+    service: str
+    interval_miles: Optional[int] = Field(default=None, alias="intervalMiles")
+    interval_months: Optional[int] = Field(default=None, alias="intervalMonths")
+    details: Optional[str] = None
+    confidence: float = Field(ge=0.0, le=1.0)
+    subtypes: list[str] = Field(default_factory=list)
+
+    model_config = {"populate_by_name": True}
+
+
+class ManualExtractionResponse(BaseModel):
+    """Response from manual extraction endpoint."""
+
+    success: bool
+    vehicle_info: Optional[ManualVehicleInfo] = Field(default=None, alias="vehicleInfo")
+    maintenance_schedules: list[ManualMaintenanceSchedule] = Field(
+        default_factory=list, alias="maintenanceSchedules"
+    )
+    raw_tables: list[dict] = Field(default_factory=list, alias="rawTables")
+    processing_time_ms: int = Field(alias="processingTimeMs")
+    total_pages: int = Field(alias="totalPages")
+    pages_processed: int = Field(alias="pagesProcessed")
+    error: Optional[str] = None
+
+    model_config = {"populate_by_name": True}
+
+
+class ManualJobResponse(BaseModel):
+    """Response for async manual extraction job."""
+
+    job_id: str = Field(alias="jobId")
+    status: JobStatus
+    progress: Optional[int] = Field(default=None, ge=0, le=100)
+    estimated_seconds: Optional[int] = Field(default=None, alias="estimatedSeconds")
+    result: Optional[ManualExtractionResponse] = None
+    error: Optional[str] = None
+
+    model_config = {"populate_by_name": True}