feat(classify): add rule-based document classification engine, task, and API

Implements the classify pipeline step with:
- Classification rules engine (app/utils/classification_rules.py) with
  pre-built categories (invoice, contract, receipt, letter, report,
  bank_statement, tax_document, insurance, payslip) and support for
  filename patterns, content keywords, and metadata matching rules
- Celery task (app/tasks/classify_document.py) that runs as a pipeline step
- CRUD API (app/api/classification_rules.py) for managing custom rules
- ClassificationRuleModel in app/models.py with migration 027
- Updated pipeline step config_schema and stage mapping
- Comprehensive tests for engine, API, and task

Co-authored-by: christianlouis <361235+christianlouis@users.noreply.github.com>

app/api/pipelines.py

@@ -117,8 +117,14 @@ PIPELINE_STEP_TYPES: dict[str, dict[str, Any]] = {
     },
     "classify": {
         "label": "Document Classification",
-        "description": "Classify the document type using AI without full metadata extraction.",
-        "config_schema": {},
+        "description": "Classify the document type using built-in and custom rules (filename patterns, content keywords, metadata matching).",
+        "config_schema": {
+            "use_builtin_rules": {
+                "type": "boolean",
+                "default": True,
+                "description": "Include the pre-built classification rules (invoice, contract, receipt, etc.).",
+            },
+        },
     },
 }