Improve and simplify post_fail_hook existence checks
    
    Summary:
    The current installedtest post_fail_hook assumes /var/tmp/abrt
    exists at all, and dies if it doesn't, leading to no /var/log
    upload. We can also avoid using openQA `script_output` - which
    is annoyingly indirect and slow - by using this neat `test -n`
    trick I found on SO. Let's also use it in the anacondatest
    post_fail_hook to avoid uploading /var/tmp when it's empty
    (which we currently do). This also drops the 0 arg from a few
    more script_run calls, because it's safe to wait for the run
    to complete and we should probably do so to avoid later typing
    errors if the commands are slow.
    
    Test Plan:
    Cause both anaconda and installed tests to fail and
    check the hooks work as intended. Maybe twiddle the failures to
    ensure directories do and don't exist and/or have contents and
    make sure things work OK. I've tested this to some degree and
    I'm pretty sure it works right.
    
    Reviewers: jskladan, garretraziel
    
    Reviewed By: garretraziel
    
    Subscribers: tflink
    
    Differential Revision: https://phab.qadevel.cloud.fedoraproject.org/D1041