Add the option to dry run every command and ensure that the command is correct

[remyleone]

It would be helpful to have a way to check that a command passed is correct from an argument parsing and typing point of view, without executing it.
It would be helpful in the case of training LLM to generate CLI commands.