v0.18.1-rc0: cmd/launch: skip --install-daemon when systemd is unavailable (#14883)

Ollama, the popular open-source platform for running large language models (LLMs) locally, has pushed a new release candidate, v0.18.1-rc0, to address a significant deployment hurdle. The core fix resolves a bug where the command `ollama launch openclaw` would fail in container environments like Docker and Kubernetes. The failure occurred because the launch script attempted to pass an `--install-daemon` flag to create a systemd user service, which is unavailable in most containers. This caused the process to exit with an error, aborting the model launch entirely, even though the model's gateway process could have run successfully in the foreground.

The update introduces a smarter check: the `--install-daemon` flag is now only passed when the system is confirmed to have systemd user services available. This is determined by checking for the presence of `/run/systemd/system` and the `XDG_RUNTIME_DIR` environment variable on Linux. On all other platforms, including macOS, Windows, and containerized Linux without systemd, the flag is omitted. This change is crucial for developers and DevOps engineers who rely on Ollama to serve models like OpenClaw within scalable, ephemeral container environments, a common pattern in CI/CD pipelines and microservices architectures.

While a seemingly technical patch, this fix removes a major point of friction for deploying Ollama in production-like scenarios. It underscores the project's maturation as it adapts to the realities of modern software deployment, moving beyond just a local developer tool to a more robust service component. The release candidate status allows the community to test the fix before it's rolled into a stable version, ensuring broader compatibility for the platform's 165k GitHub stars.