Migrate database setup from sync to async SQLAlchemy

[szyszkapiotr]

Related issue: #97

Description

This PR intruduces change from existing synchronous SQLAlchemy setup to its async (asyncpg) version and extends the generic CRUD/service layer with bulk and lookup helpers. Additionally, a commit: bool flag has been introduced to repository methods (default: True). This allows callers that compose multiple repository operations to pass commit=False and let the service layer own the transaction, committing only once at the end.

Changes

Async migration

• config.py.jinja: add db_async_uri property (postgresql+asyncpg://).
• database.py: replace sync create_engine/sessionmaker/Session with
  create_async_engine / async_sessionmaker / AsyncSession. Export
  AsyncDbSession dependency, adb_session_ctx, and AsyncSessionLocal
  (expire_on_commit=False). Drop engine, SessionLocal, DbSession.
• repositories.py, services/services.py: convert all CRUD/service methods to
  async/await; queries moved from session.query(...) to select()/delete().
• user/repositories/activity_repository.py, user/services/activity_mixin.py:
  migrate is_user_active to async select().
• utils/healthcheck.py: async db.execute, pool status via async_engine.
• utils/exceptions.py: handle_exceptions now branches on
  asyncio.iscoroutinefunction so exception-to-HTTP translation works for async
  methods (previously the sync wrapper returned the coroutine unawaited and never
  caught anything).

New repository/service methods

• create_many — bulk insert; flushes when commit=False.
• get_from_ids — fetch a set of rows by id list.
• get_filtered_scalar — single-or-none lookup, raises MultipleResultsFoundError when more than one row matches.
• delete_many — batched bulk delete; deletes per batch then commits once
  (atomic across batches).
• commit flag added to create/update/delete for unit-of-work composition.