update nuclear capacities

[pweigmann]

Issue https://github.com/remindmodel/development_issues/issues/555

This is an update of the nuclear capacity data, which is fed into REMIND and serves as historical reference and near-term upper bounds. The corresponding REMIND PR to adjust core/bounds is not ready yet, but of course the two will be merged together.

Working on this was a bit confusing, as it seems the file was created and updated by different people over the last 9 years (at least) and some comments/descriptions were not adjusted accordingly. Also, it might be a bit dangerous to mix Capacity and Capacity Additions within one data object with only one variable and trust people read the descriptions. I cleaned up the code and clarified some descriptions, hopefully without adding new misunderstandings. @robertpietzcker please take a look at the result to spot any incorrect approaches or assumptions in the new version.

In detail

• Taking a current snapshot of the nuclear data from here and feeding it into calcCapacityNuclear
• 2015, 2020 and 2025 are now directly using the capacities from the respective snapshots.
• For capacity additions in 2030 and 2035 I used the formulas as were used for 2025 and 2030 before and made small changes.
  • The "10% of operating" are taken from the same (latest) snapshot instead of an older one. I did not see why it would make sense to take an older one.
  • Are the "30% under construction" for 2030 really the right fraction? Why do we expect 40% of planned but only 30% of under construction in 5 years? I did not change it, but feel like this should be bigger.
  • I also updated the manual adjustments made in the end for emerging nuclear energy countries. Should we continue with the same estimates for capacity additions of 500MW in 2030 and 2GW in 2035 as before?

We are now the same place we were with CCS, where we have REMIND input-data and bounds in parallel to near-term thresholds used in the validation tool, both based on the same source but with slightly different factors and methods. I don't think it should be harmonized, but it might be good to keep this in mind if only to avoid double work.