I see two use cases for this (for continuous deployment purposes):
- Initial installation of the software can be automated. This feels a bit cleaner than requiring the "first" install to be user-interactive, because it is repeatable.
- If a new component is added which is checked by default, the command line provides an extra level of control of where that gets installed. In fact, without it, I imagine it would automatically be installed - you'd need to stop the automated process, and go manually run the installer, uncheck the component, and then re-enable everything to prevent that new component from being automatically installed.