Split-extract retrieve_and_process_data into two methods

As its name was implying, that method was doing two different things:
first retrieving the raw data − ie the groupings and the column data,
and processing that data into percentage and formatting it as wikitable.

The first part was particularly nasty because it was all done as side-effect:
the column_data class attribute was simply populated on the go.

Here we separate the concerns between the two, and add the most
trivial of unit tests to validate the behaviour.

Let's not invest more time into good tests as this a stepping
stone to the major Line/Grouping refactoring.