perf(db): add composite indexes for high-frequency financial queries

[bkennedyshit]

Closes #128

What

Adds 6 composite SQLAlchemy indexes targeting the most common query patterns in FinMind's financial data layer.

Indexes Added

Index	Table	Columns	Query Pattern
ix_expenses_user_spent	expenses	user_id, spent_at	Dashboard date-range aggregations
ix_expenses_user_category	expenses	user_id, category_id	Category breakdown reports
ix_expenses_user_type	expenses	user_id, expense_type	Income vs expense splits
ix_recurring_user_active	recurring_expenses	user_id, active	Active recurring expense fetches
ix_bills_user_due	bills	user_id, due_date	Upcoming bills lookups
ix_categories_user	categories	user_id	Per-user category list

Implementation

• Alembic migration 003_add_performance_indexes.py with upgrade() and downgrade()
• models_indexed.py — ORM-level Index() definitions for SQLAlchemy awareness
• Tests — verify index definitions exist and queries execute without error

Testing

pytest packages/backend/tests/test_db_indexes.py -v