Improve error handling and add comprehensive client tests

[iMicknl]

Summary

• Fix get_current_execution crashing on empty responses (cloud returns {}, local returns [] or null)
• Make ActionGroup.id and ActionGroup.oid optional (str | None) — inline action groups from exec/current don't carry identifiers
• Add start_time, execution_type, and execution_sub_type fields to Execution model (matching real API responses)
• Change Execution.state from str to ExecutionState enum (consistent with HistoryExecution)
• Add NoSuchDeviceError and NoSuchActionGroupError exception types with response handler mappings
• Add 8 new cloud exception fixture tests and 6 new cloud exception JSON fixtures
• Add 13 cloud client tests covering executions, history, state, places, action groups, and event listeners
• Add 10 local API tests verifying proper error handling for KizOs gateway differences

Breaking changes

• Execution.state is now ExecutionState instead of str, ActionGroup.id and ActionGroup.oid are now str | None instead of str, get_current_execution now returns Execution | None instead of Execution

Test plan

• All 311 unit tests pass
• All prek checks pass (ruff, mypy, ty)
• Validated against real Somfy cloud API (37/37 probes pass)
• Validated against real local KizOs gateway (37/37 probes pass)
• JSON fixtures verified against real API response structures

[Copilot]

Pull request overview

Improves robustness of Overkiz client error/edge-case handling (notably executions and missing resources), aligns execution parsing with real API payloads, and adds broad unit-test coverage plus a real-API probing script.

Changes:

• Make get_current_execution() resilient to empty/invalid “not found” payloads by returning None.
• Extend/align models and error mapping: Execution gains timing/type fields and uses ExecutionState; add NoSuchDeviceError / NoSuchActionGroupError; allow inline ActionGroup objects without identifiers.
• Add many new unit tests + JSON fixtures for cloud/local differences, plus a probe script to validate error handling against real endpoints.

Reviewed changes

Copilot reviewed 20 out of 20 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
pyoverkiz/client.py	Returns None for empty/non-dict exec/current/{id} responses to prevent crashes.
pyoverkiz/models.py	Aligns Execution/ActionGroup parsing with observed API payloads; adds execution metadata fields and enum state.
pyoverkiz/response_handler.py	Maps NO_SUCH_DEVICE / NO_SUCH_ACTION_GROUP to new dedicated exceptions.
pyoverkiz/exceptions.py	Introduces NoSuchDeviceError and NoSuchActionGroupError.
tests/test_client.py	Adds comprehensive tests for executions, history, places, state parsing, action execution, and local-gateway error variants.
tests/fixtures/**	Adds endpoint and exception JSON fixtures used by new tests.
scripts/test_error_handling.py	Adds a probe tool to validate error handling behavior against real cloud/local APIs.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.