fix: various issues with HIFLD transmission loading/processing

[danielolsen]

Pull Request doc

Purpose

Various fixes for processing of HIFLD transmission data:

• Correct a lat/lon vs. lon/lat issue (and add a test for said issue) when mapping lines to substations
• In addition to lines whose status is IN SERVICE, also return lines whose status is NOT AVAILABLE, since this represents most of the lines in the Rocky Mountain states (see Improve handling of DC lines when loading HIFLD data #233 (comment)).
• Using a rough bounding box of the contiguous U.S., don't return lines for which at least one end is outside of this box (e.g. Alaska, Hawaii, Puerto Rico)
• Ensure that indices aren't duplicated within the branch dataframe as a results of concatenating individual dataframes for [original lines, MST-connecting lines, transformers]

What the code is doing

• In const.py, we add data for a rough bounding box for the contiguous U.S. states.
• In load.py, we use this bounding box to filter lines, and we return both IN SERVICE and NOT AVAILABLE lines.
• In transmission.py:
  • within map_lines_to_substations_using_coords, we fix a lat/lon vs. lon/lat bug.
  • within filter_islands_and_connect_with_mst, we add a meaningful index to the new lines.
  • within build_transmission, we add a meaningful index to the new transformers.
• In test_transmission.py, we add a test (test_map_lines_to_substations_using_coords) that catches the lat/lon vs. lon/lat issue as well as tests the combining of substations via coordinate rounding.

Testing

The new unit test initially fails due to the lat/lon issue, but passes once the fix is added.

Usage Example/Visuals

Before:

>>> from prereise.gather.griddata.hifld.data_process.transmission import build_transmission
>>> branch, bus, substations, dc_lines = build_transmission()
>>> len(branch)
74859
>>> branch.index.duplicated().sum()
522

After:

>>> from prereise.gather.griddata.hifld.data_process.transmission import build_transmission
>>> branch, bus, substations, dc_lines = build_transmission()
>>> len(branch)
93908
>>> branch.index.duplicated().sum()
0

After the changes, we have many more branches, and no duplicate indices.

Time estimate

30 minutes.

[BainanXia]

Thanks!