#967 takes the point of view, that the capacities added by the model in 2025 correspond to the time interval [2025,2030]. The logic of the grouping years for existing capacities was different and solar-2020 would correspond to capacities installed in [2015,2020]. #967 aligned the way the grouping years work, s.t. solar-2015 would correspond to the interval [2015,2020].

However, as pointed out in PyPSA/pypsa-ariadne#101 the interpretation that solar-2025 corresponds to [2025,2030] might be flawed. In pypsa-ariadne we impose capacities and co2 limits corresponding to actual laws. These limits apply to the year 2025, thus any capacities extensions of the solar-2025 generator required to achieve these limit have to be built before 2025. Hence the more correct point of view is that the capacities expanded by the model in 2025 correspond to the time interval [2020,2025]. This is how the grouping years of existing capacities worked before #967, and thus this PR mostly reverts #967.

Open issues:

• grouping years for heating may not extend 2020. Otherwise infeasibilities occur. I have no idea why, but i would prefer if this would be handled in the code and not in the config

• The renewable capacity added in the base year is neglected because of the land use constraint in solve_network. it would be desirable to have separate Generators for extendable and existing renewable capacities.