storage_location in its argument. It is the base
directory where the data are stored to and loaded from.
The methods load_table() and determine_field_names()
interpret the argument table_name as a subdirectory in the base directory in which each attribute is stored as a
single file in a binary format. The file names correspond to attribute names
with an extension that determines the type of the stored data. The file
extension consists of three parts, one character for the byte order, one
character for the type of the data, and one or more characters for the size of
the column in bytes. The extension is similar to the dtype for numpy arrays,
only the charakter for the byte order is changed from '<', '>', or '|' (for
little endian, big endian, or irrelevant) to 'l', 'b', or 'i'.
The method write_dataset() stores attribute according to the scheme
above. By default the file extensions are determined automatically. This can
be changed by an entry valuetypes in write_resources, which is a
dictionary with a type for each attribute.