Generate individual history tables

[wergstatt]

In the current approach implemented in #1 there is a single table that updates whenever any column is changed. A more parsimonious implementation that does only focus on single columns should be advantages and more traceable. Memory shouldn't be relevant for our 2000 journals.

• [x] As a prototype, derive a table with foreign keys to journal and publisher that does only track the evolution of the rank of the journal.
• [x] Remove the current implementation, which spams a little bit.
• [x] In a second step we can think of tracking other statistics as well (e.g. impact factor or number of articles).

[wergstatt]

Now we've got tables for

• Number of articles
• Impact Factor
• Rank