Fix ml-service memory leak

[myungjoo]

NOTE: This does not fix the client's reported bug.

Fixes a memory leak by ensuring proper pipeline destruction and state management in ml-api-service-agent-client.

The previous implementation could lead to memory leaks on the daemon side when ml_service_pipeline_launch() and _ml_service_pipeline_release_internal() were called repeatedly from multiple threads or processes. This was due to server->id not being correctly initialized to indicate an unlaunched state and not properly handling pipeline ID 0 during destruction, or not resetting the ID after destruction, leading to orphaned pipelines.

---

 

[cursor]

Cursor Agent can help with this pull request. Just @cursor in comments and I'll start working on changes in this branch.
_{Learn more about Cursor Agents}