chore(liblogosdelivery): node_api destroy move + security/resource fixes

[NagyZoltanPeter]

Summary

• move destroy api to node_api
• add security checks
• fix possible resource leak

Scope after split

• this PR now intentionally excludes health-status event support
• health-status event support was extracted to: feat(liblogosdelivery): health status event support #3737

[Copilot]

Pull request overview

This PR addresses security concerns in the FFI API and adds health status monitoring capabilities to liblogosdelivery. The changes improve resource management, add partial race condition protection for event callbacks, and enable connection status monitoring through the FFI interface.

Changes:

• Added Lock-based protection for event callback setter to mitigate race conditions
• Implemented resource cleanup when FFI thread request submission fails during node creation
• Added EventConnectionStatusChange listener to expose health status events to FFI consumers

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 6 comments.

File	Description
liblogosdelivery/liblogosdelivery.nim	Removed logosdelivery_destroy function (relocated to node_api.nim)
liblogosdelivery/declare_lib.nim	Added Lock import and lock-protected event callback setter to prevent race conditions
liblogosdelivery/logos_delivery_api/node_api.nim	Added logosdelivery_destroy function, health event listener, resource cleanup on creation failure, and event listener cleanup in stop

---

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

[Copilot]

If event listener registration fails (e.g., MessageSentEvent.listen succeeds but MessageErrorEvent.listen fails), the function returns early without cleaning up the previously registered listeners. This could lead to orphaned event listeners that continue to trigger callbacks even though the node never fully started. Consider adding cleanup logic in the error paths, or using a defer block with a success flag to ensure listeners are cleaned up on failure.

[Copilot]

Spelling error in comment: "due incorrect usage" should be "due to incorrect usage".

Suggested change

	# prevent race conditions that might happen due incorrect usage.
	# prevent race conditions that might happen due to incorrect usage.

[Copilot]

The variable name ConnectionStatusChangeListener should follow camelCase convention to be consistent with the other listener variables (sentListener, errorListener, propagatedListener) in the same function. Consider renaming it to connectionStatusChangeListener.

Suggested change

	let ConnectionStatusChangeListener = EventConnectionStatusChange.listen(
	let connectionStatusChangeListener = EventConnectionStatusChange.listen(

[Copilot]

The listener variables (sentListener, errorListener, propagatedListener, ConnectionStatusChangeListener) are assigned but never used. While this may be intentional if the listeners are automatically managed by the event broker system, consider whether these variables should be stored or if they can be ignored with an underscore prefix (e.g., _ = MessageSentEvent.listen(...)). If they need to be stored for cleanup, they should be saved in the context. Note that cleanup currently uses dropAllListeners which doesn't require these handles.

[Copilot]

Event listener cleanup in logosdelivery_stop_node should ideally happen after the node is stopped, not before. If the node emits events during shutdown, those events won't be properly handled since the listeners have already been dropped. Consider moving the dropAllListeners calls after the ctx.myLib[].stop() call, or handle this within the stop implementation itself.

[fcecin]

LGTM!