Backport robustness improvements from python-s7comm

[gijzelaerr]

Summary

Backports concrete robustness improvements from nikteliy/python-s7comm to the native Python S7 implementation:

• 210+ S7 protocol error codes from Wireshark's S7 dissector — covers USERDATA parameter errors (0xD0xx), protocol errors (0x8xxx), resource errors, and more. parse_response() now raises descriptive S7ProtocolError on header errors, and USERDATA response parsing logs descriptive warnings.
• Stale packet detection with retry — new validate_pdu_reference() validates response sequence numbers. New _send_receive() helper wraps the send→receive→parse pattern with automatic retry on stale packets (up to 3 retries). Refactored ~15 call sites in client.py to use it.
• Automatic PDU splitting for read_area()/write_area() — requests exceeding the negotiated PDU length are automatically chunked, preventing PLC rejections on large reads (e.g., db_get() on big DBs).
• TPDUSize enum (ISO 8073) with configurable COTP negotiation — replaces hardcoded 0x0A with a proper TPDUSize IntEnum (128–8192 bytes).
• MAX_VARS=20 limit on read_multi_vars()/write_multi_vars() — raises ValueError when exceeding the S7 protocol limit.

Not incorporated: async client, Protocol-based typing, packet-per-class architecture, PCAP-based testing.

Test plan

• 27 new unit tests in tests/test_improvements.py covering all features
• All 428 tests pass (401 existing + 27 new)
• ruff check, ruff format --check, mypy all pass clean

🤖 Generated with Claude Code