[P1/quality] Later pycubrid recipes (07-16) quality regression — missing annotations, getattr hack, missing imports

[yeongseon]

Description

Recipes 01-06 in fundamentals/pycubrid/ are high quality (full type annotations, proper imports, clean patterns). Recipes 07-16 show a significant quality drop:

1. Missing from __future__ import annotations

• 07_merge_upsert.py, 08_hierarchy_connect_by.py, 09_serial_order_numbers.py, 10_collection_columns.py
• Required by project AGENTS.md

2. getattr(pycubrid, "connect") anti-pattern

• 11_bulk_etl_pipeline.py:21, 12_pool_retry_worker.py:18, 13_atomic_counters.py:21, 14_manual_cascade_delete.py:16
• Uses CONNECT = getattr(pycubrid, "connect") instead of direct pycubrid.connect
• Obscures the API for learners and disables IDE autocomplete

# Current (bad)
import pycubrid
CONNECT = getattr(pycubrid, "connect")  # ← confusing for learners

# Should be
import pycubrid
# ... use pycubrid.connect() directly

3. Missing type annotations

• Functions in 07-14 lack return type annotations
• Earlier files (01-06) are fully annotated

4. File-level # pyright: reportAttributeAccessIssue=false

• Files 07-12 suppress pyright globally
• Masks real type errors

5. Import ordering violations

• from datetime import datetime placed after import pycubrid (stdlib should come first)

Fix

1. Add from __future__ import annotations to files 07-10
2. Replace getattr(pycubrid, "connect") with direct pycubrid.connect()
3. Add type annotations to match quality of files 01-06
4. Replace file-level pyright suppression with inline # type: ignore[attr-defined]
5. Fix import ordering: stdlib → third-party → local

Context

Found during line-by-line code review (2025-07-23).

[yeongseon]

Fixed in PR #34 (merged).