When calling environment new, the CLI can TCP-timeout and exit with an error — but Railway has already created the environment on the backend. From the caller's perspective the request failed, but retrying with the same name returns "already exists."

This makes it impossible to safely retry environment creation in CI pipelines: we can't tell whether to retry (env wasn't created) or clean up (env was created silently).

We'd love environment new to be idempotent — if an environment with the given name already exists and was created within the same request window, return success instead of an error. Alternatively, a way to detect "creation in progress" vs "already exists from a prior run" would help.