More ways of running the tests

[hrj]

We are currently comparing the screenshot of the test with the screenshot of the reference. This assumes that the references are being rendered correctly.

However, there might be bugs in rendering the references itself. And these bugs might come and go as the browser evolves. To keep track of these, we need to store baseline screenshots. The reference screenshots can be then compared, between the current version of the browser and the baseline version of the browser.

I propose these commands:

• prepare
• capture : to capture screenshots
• compare : to compare test with reference screenshots
• compare-baseline <base-directory> : to compare reference screenshots between two browser versions

[hrj]

The comparisons are now being done through report.html, which is good enough for now.