Improve the metadata collection function

[gutzbenj]

The German Weather Service has stored his metadata in a poor format, which has no seperators (neither comma nor tab). Instead of reading the data like we currently do it (rowwise reading and checking of number of chunks, then processing it with respect to this chunk number),
`

from fix_metaindex

metaindex_to_fix = metaindex.iloc[:, 6:]

# Reduce the original dataframe by those columns
metaindex = metaindex.iloc[:, :6]

# Index is fixed by string operations (put together all except the last
# string which refers to state)
metaindex_to_fix = metaindex_to_fix \
                       .agg(lambda data: [string
                                          for string in data
                                          if string is not None], 1) \
                       .to_frame() \
                       .iloc[:, 0] \
    .agg(lambda data: [' '.join(data[:-1]), data[-1]]) \
    .apply(pd.Series)

# Finally put together again the original frame and the fixed data
metaindex = pd.concat([metaindex, metaindex_to_fix], axis=1)

`

we should find another way, where we can read it directly with pandas with some predefinitions like read_fwf (fixed width formatted). This should give as a performance boost and also fix some other issues with string handling.