In pypsa==0.25.2, the clustering strategy is 'consense' per default.
That created problems for some columns, which are solved by adding specific strategies for these columns.

From pypsa>0.25.1 on, the aggregation of already solved networks creates problems in pypsa.
It looks for all time series variables in the static dataframe, which leads to key errors, eg. for 'p' of generators.
This commit aggeragted the new generators using pandas functions.

Should be removes/replaced later on

Within pypsa's network clustering, all output timeseries are filled with 0.0.
This results into problems in the following clustering steps.
To avoid that, the output timeseries data is replaced with empty dataframes.

PyPSA's networkclustering now deletes timeseries that only inlcude the default value.
When running smaller example calculations eith only a few snapshots, it could
happen, that the p_max_pu time series of some wind or solar generator is always 1.0
To avoid key errors in these cases, it was required to not use the carrier to
decide if a generator has a p_max_pu time series but directly look at the existing columns.

…uded

Before it was checked if the last snashot of the disaggreagted timeseries is
included which led to duplicated index values.