Keep PSS/E HVDC impedances in ohms

[m-bossart]

PSS/E cases with HVDC lines imported into PowerSystems with LCC and VSC impedances per-unitized twice.
psse.jl converted these impedances to per-unit, dc_branch.jl tagged the result as ohms ("ohm") or siemens ("S"), and the PowerSystems importer converted them again.
LCC resistances and reactances came out too small by a factor of Z_base, and VSC g was off by the same kind of factor.
Power flows on any PSS/E case with HVDC lines were wrong.

Changes

• psse.jl keeps RDC, RCR, XCR, RCI, XCI, XCAPR, XCAPI and the VSC RDC in ohms, which is the unit the OpenAPI writer expects.
  compounding_resistance was already in ohms, so all LCC impedances now use the same unit.
• An out-of-service LCC line (MDC=0) with a zero scheduled DC voltage (VSCHD) now parses without a warning, since the parser no longer converts to per-unit.
  An in-service line with VSCHD=0 still raises an error.

dc_branch.jl and PowerSystems need no change.

Testing

• New tests compare the parsed data and the OpenAPI values against the raw PSS/E fields for LCC and VSC lines.
  They failed before this change.
• The full test suite passes.
• End to end (parse, OpenAPI document, PSY.from_file), every LCC and VSC impedance matches a hand calculation from the raw file on two-terminal-hvdc_test, vsc-hvdc_test, case16_all_components and
  case25_v35_savnwb.

Known remaining issue

For an out-of-service LCC line with VSCHD=0, the PowerSystems importer still divides by zero and sets r to Inf.
That already happened before this change, and it needs a fix in PowerSystems.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!