- Arguments with spaces like foo:'bar baz' were improperly passed to
Taskwarrior before. This only affected bash_tap.sh tests, the Python
tests have always been handling this properly.
- Now that bash_tap.sh no longer uses subshells, TASKDATA can and
and should be used. This fixes a bug introduced with the switch
to "eval" where having TASKDATA set while running tests would
potentially lead to problems.