Geodata features are not properly documented

[khaeru]

186 (superseding #165) added methods TimeSeries.add_geodata, TimeSeries.geodata, etc. These are not fully documented.

• Reading the PR (i.e. on GitHub), one sees the text: “Geodata here means meta-information about layers stored in a geoserver.”
• Reading the test suite, one can see that the value column stored/returned by these methods is a string, rather than a float: https://github.com/iiasa/ixmp/blob/1c3fc190281ea7588ea660ee883fb762bff1cfe4/tests/test_feature_geodata.py#L7-L16

However, a new user who reads the actual documentation will not see either of these clues, and so will be mystified about:

• what “geodata” specifically means in the context of ixmp,
• how to use it (i.e. what values they can, and can not, pass to these methods),
• what should be returned, and
• anything that is done by the Java backend to manipulate this data.

At a minimum, these things need to be added to the method docstrings.

[khaeru]

I updated #182 to incorporate these methods.

The necessary code reveals that the signatures and return values of the Java get/set/delete methods for ordinary (float) timeseries data, and timeseries geodata, are quite different, even though (per above) the only substantive difference seems to be the type of the 'value' column.

If these discrepancies could be cleaned up on the Java side, then the JDBCBackend could also be simplified.