So far we have only focused on a very limited set of valid Markua tags. As we make progress we'll need to be able to report rows in an input file that we could not recognize. So let's add that error collecting code now. It will make our parsing tighter as it will mean that for every line in the input file we'll have to know exactly what it means.

We can already foresee some trouble with our current examples as we have not handled the empty rows at all.