Overview of changes:

 *  Applied proper HTML escaping

    Ie. HTML content and similar should show up properly in messages et
    al.

 *  Switched message passing in translator functions to via stdin

    Just like embedded data is already passed, pasing via stdin avoids
    the problem that Jinja2 can't escape for Bash.

 *  Ensure YAML is always read as utf-8

 *  Fixed crash when adding lint to EventTree itself

    For example, missing phase end could trigger this.

 *  Codebase maintenance

Passing via argument is bound to fail on quoting issues

Overview of changes:

 *  Fix jinja2 requirement for RHEL-7

    Turns out that RHEL-7 also calls it python2-jinja2

 *  Updated MKit for test suites to v0.0.35

Overview of changes:

 *  Added Jinja2 template for beakerlib translate script

    Generates shell script that can be called from within beakerlib test
    to "re-enact" the JAT in beakerlib language.

 *  Updated log data model with session and test identifiers

 *  Added test id to HTML report title

 *  Added missing PyYAML dependency

 *  Codebase maintenance

We now depend on jat 0.0.2

Overview of changes:

 *  Fixed dependency tree

Mostly for debian/control, which has been a bit neglected for a time.

Overview of changes:

 *  Set release source and destination branches in skeleton

 *  Added debugging around test execution

Overview of changes:

 *  Support jattool-report (and whole jattool) on older distros

    It was assumed that full jattool won't be available on RHEL-6 and
    CentOS 6 due to Jinja, but that's not correct.

 *  Add mention of convenience script for enabling repos on old boxes.

The assumption that RHEL-6 and older don't have Jinja2 was incorrect;
just the package is named differently.  However, in many cases people
will still want small installs, so we're keeping the split, just the
note is rephrased to be more accurate.

RHEL6 and older call it python-jinja2, then it's called python2-jinja2.

Fixes #3.

Overview of changes:

 *  Fixed test discovery code

    Most of the code was broken; discovery by id, version printing...
    Now it's still embarrassingly slow, but at least it somewhat works.

 *  Updated MKit for jattool and test suites to v0.0.34

 *  Updated parser, Jinja templats and more for newer shellfu-bash-jat

 *  Added meaningful content to README.md

 *  Added support for showing test/suite version

 *  Changed 'jats' term to 'id' for jats:// test and suite ids

 *  Ensured proper test status from runtest matches qrun (and

 *  Split-off jattool-minimal for testing machines

 *  Changed jats: ids to use double-slash between suite part and test part

 *  Added error message if test has not finished cleanly

 *  Rewrote broken parts of test discovery and identification

 *  Codebase maintenance

Not very useful alone, but already good to read version of a test suite
or test, given that path is passed as well.

Needless to say, the HTML template is still horrible and needs someone
with better HTML/CSS skills.

The Python parser is still not really correct, but at least does not
depend on function names.

Keep jattool as meta-package for easy installations on workstations.

There's no way this would have worked properly; I don't know what I was
thinking of.