Get Mariona's algorithms up and running

[alexhebing]

Continues from #26. Client choses to go for 1), but wants to see if the algorithm actually performs.
Do minimal updates and try to get the scripts running as they are, so that some basic tests are possible.

[alexhebing]

I got the script up and running by updating to Python 3.4. Most changed were needed to the database connection (handled PyMySQL now instead of MySqlDb). The script do include some very dirty fixes now (all with comments), simply because I wanted them to run.

Tested with the Dutch corpora included with the scripts, and a new (!), tiny corpus, NERred by Stanford (and manually edited to remove 'B-' and 'I-' from the tags), and it actually works!

Will take these results with me to client and see if / how we can establish if the algorithm performs well. (thinking of looking at the texts and the coordinates found)