fix: move NullPool test fix from db.py into a conftest-local test engine

Production code shouldn't branch on `"pytest" in sys.modules` — it's
fragile and couples db.py to test tooling. Instead, conftest.py now
builds its own dedicated NullPool engine directly from settings, used
only for the drop_all/create_all reset and the get_db override. The
production engine in app/db.py is untouched and never exercised during
tests, so this fully preserves the event-loop fix while keeping prod
code test-agnostic.

backend/app/db.py

@@ -1,23 +1,9 @@
-import sys
-
 from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker, create_async_engine
 from sqlalchemy.orm import DeclarativeBase
-from sqlalchemy.pool import NullPool
 
 from app.config import settings
 
-# pytest-asyncio gives each test function its own event loop by default, but
-# this engine (and its connection pool) is a module-level singleton created
-# once and reused across every test. asyncpg connections are bound to the
-# event loop they were opened on, so pooling them across per-test loops
-# raises "got Future attached to a different loop". NullPool sidesteps this
-# by opening a fresh connection per checkout instead of reusing one from a
-# prior loop. Only applied under pytest; production keeps normal pooling.
-_engine_kwargs = {"echo": False}
-if "pytest" in sys.modules:
-    _engine_kwargs["poolclass"] = NullPool
-
-engine = create_async_engine(settings.database_url, **_engine_kwargs)
+engine = create_async_engine(settings.database_url, echo=False)
 async_session_maker = async_sessionmaker(engine, expire_on_commit=False)