feat: Bank Sync Connector Architecture ($500 bounty)

[jeromeyhq]

Bank Sync Connector Architecture

Pluggable architecture for bank integrations as specified in #75.

What this delivers

• Connector Interface — Abstract BankConnector base class with connect(), disconnect(), import_transactions(), and refresh() methods
• Import & Refresh Support — Bulk import with date range + incremental refresh since last sync
• Mock Connector Included — Realistic test data across 3 accounts (checking, savings, credit) with 5 spending categories + income

Files added

File	Purpose
app/services/connectors/base.py	Abstract BankConnector interface + TransactionRecord, AccountInfo, SyncResult dataclasses
app/services/connectors/mock_connector.py	Full mock implementation with deterministic transaction generation
app/services/connectors/registry.py	Pluggable registry with register_connector(), get_connector(), list_connectors()
app/routes/bank_sync.py	REST API: /api/bank-sync/{connectors,accounts,import,refresh,health}
tests/test_connectors.py	Full test suite (pytest)

API Endpoints

GET  /api/bank-sync/connectors   — List available connectors
GET  /api/bank-sync/accounts     — List bank accounts for a provider
POST /api/bank-sync/import       — Bulk import transactions
POST /api/bank-sync/refresh      — Incremental refresh
GET  /api/bank-sync/health       — Health check

Usage example

# List connectors
curl http://localhost:5000/api/bank-sync/connectors

# Import transactions
curl -X POST http://localhost:5000/api/bank-sync/import \
  -H "Content-Type: application/json" \
  -d '{"provider": "mock", "account_id": "mock-checking-001"}'

# Refresh
curl -X POST http://localhost:5000/api/bank-sync/refresh \
  -H "Content-Type: application/json" \
  -d '{"provider": "mock", "account_id": "mock-checking-001"}'

Design notes

• Python 3.9 compatible (no 3.10+ type syntax)
• Easy to add new connectors: just inherit from BankConnector and use @register_connector
• Mock connector generates deterministic data (same date range = same transactions)
• Transaction records normalize to FinMind's existing expense import format

All 5 Python files compile cleanly. Tests included.