Add a framework for handling hierchies of field types. Each field type can have a field name, a parent type, a maximal occurrence count (not yet enforced) and a data type (currently string, binary, float or int). In addition to this, the field type system will have the possibility for the indexwriter to store arbitrary data. A potential use for this is the ablity to cache fieldnames in a way that can be written faster.

This is, for example, useful when using a mapping from one fieldname to another. By storing the name to be mapped to, expensive lookups can be avoided.