Select aggregation functions based on column types

[terror]

Resolves #26

Putting this up for visibility. This MR allows us to generate groupby + aggregations of the form:

.groupby(by=['S_SUPPKEY', 'PS_PARTKEY', 'BRAND', 'N_REGIONKEY', 'PS_SUPPKEY']).agg({'S_NATIONKEY': 'last', 'R_REGIONKEY': 'min'})

...where the aggregation function is compatible with the column type.

There's an issue however where pd.eval throws this error when executing queries of this type:

NotImplementedError: 'Dict' nodes are not implemented

Need to find a way around this in order to get this in.