node: reap the stdio bridge backend after killing it

[aojea]

baseService.Teardown killed the command backend's process but nothing ever called Wait, so every torn-down backend stayed a zombie until the node exited. The bridge's stdout reader goroutine now reaps it: it is the only reader of the pipe, so Wait cannot race a read on the pipe it closes, and every exit path (natural exit, crash, Teardown) ends there with an EOF. Pending callers are released before Wait runs so a slow kernel teardown of the child never holds up a request.

Teardown treats os.ErrProcessDone as success, since a backend that exited on its own has already been reaped by the time it is torn down.

[gemini-code-assist]

Code Review

This pull request improves process lifecycle management and cleanup in the node service. Specifically, it ensures that killed command backends are properly reaped by calling Wait() in the StdioBridge to prevent zombie processes, and updates Teardown to ignore os.ErrProcessDone errors when a process has already exited. A new unit test has been added to verify this reaping behavior. No review comments were provided, and the changes align with the repository's testing guidelines.