lisa: Use inspect.getdoc() instead of textwrap.dedent

inspect.getdoc()/cleandoc() are better suited to handle docstrings than
textwrap.dedent(), since they handle properly lack of indentation on the
first line.