Add "dedent" filter to Jinja2 environment

Kpet-db templates often have to specify ks_append element contents, and that seems to require that the text has no indentation. This requires breaking block indentation in these cases, which hurts readability, possibly resulting in issues like two "ks_appends" being defined in some tree templates.

Add the textwrap.dedent function as a filter to the Jinja2 environment to help mitigate the issue. It could then be used like this:

{% set ks_append | dedent %}
    <ks_append><![CDATA[
    %post
    set -x
    LVNAME=$(lvs -o lv_name,vg_name --noheadings | awk '/root/ {print $1}')
    VGNAME=$(lvs -o lv_name,vg_name --noheadings | awk '/root/ {print $NF}')
    lvextend -l +100%FREE /dev/${VGNAME}/${LVNAME}
    xfs_growfs /
    %end
    ]]></ks_append>
{% endset %}