Spike: statistics

In this branch I experimented with statistics. Major changes:

• The attributes used for statistics can be defined for each study (only attributes with privacy_level_read=0 as before)
• There can be more than two such attributes
• No combined charts; Instead, there is a separate chart for each attribute
• The recruitment list can be sorted by each of the attributes individually
• Display stats also in recruitment detail and execution list/detail
• Change styling of stats and also include label (see #54 (closed))
• Allow to export CSV with pseudonyms and attributes for every participation
• The AttributeField interface was simplified a bit.
• It is now possible to use any attribute type for statistics, even if it does not provide rough k-anonymity. This either has to be changed or we need to implement clustering for all types.