fix: correct generation of duplicated bus entries in bus-to-NEEM-region mapping

[danielolsen]

Pull Request doc

Purpose

Closes #545.

What the code is doing

• Within create_mapping_files.py, after we detect bus coordinates that lay within region polygons, we identify duplicated buses. For each, we find the closest neighboring bus within non-duplicated buses. We assume that that is the best guess for the 'true' region of the duplicated bus, and mark any other regions to be dropped.
• There is also an unrelated fix for the code to make directories.
• We update the CSV as a result of running the new code.

Testing

Tested manually to generate a new CSV:

from powersimdata import Grid
from powersimdata.design.investment.create_mapping_files import write_bus_neem_map
grid = Grid("USA")
write_bus_neem_map(grid)

Confirming that the new CSV has no duplicates:

>>> from powersimdata.design.investment import const
>>> mapping = pd.read_csv(const.bus_neem_regions_path)
>>> mapping.index.duplicated().sum()
0

We can also see that the newly-generated CSV now has exactly nine fewer rows, and no other changes, which follows from the nine duplicate pairs detected earlier.

Confirming that the Scenario that triggered the bug before no longer does:

>>> from powersimdata import Scenario
>>> from powersimdata.design.investment.investment_costs import calculate_ac_inv_costs
>>> scenario = Scenario(3287)
>>> calculate_ac_inv_costs(scenario)
{'line_cost': 387768866295.44775, 'transformer_cost': 9334839565.245375}

Time estimate

15-30 minutes.

[BainanXia]

Good catch.