FIX: correct default Earth radius (10x too large) in geopotential height conversions

[zjk199506]

Summary

The default Earth radius of the three geopotential conversion helpers in rocketpy/tools.py is 63781370 m — 10x the WGS-84 semi-major axis — directly contradicting their own docstrings:

def geopotential_height_to_geometric_height(geopotential_height, radius=63781370.0):
    """...
    radius : float, optional
        The Earth's radius in meters, defaulting to 6378137.0.   # <- docstring disagrees with signature

geopotential_to_height_asl and geopotential_to_height_agl carry the same radius=63781370 default, and the docstring examples of all three functions are anchored to the 10x-radius values, which masked the bug.

Impact

• The main Environment processing paths already pass self.earth_radius explicitly and are unaffected.
• The default bites when the helpers are reused directly, e.g. EnvironmentAnalysis computes surface elevation from NOAA reanalysis surface geopotential (z) via geopotential_to_height_asl(surface_geopotential) with the default radius (environment_analysis.py:876).
• Height bias from the 10x radius: ~-14 m at 10 km, ~-57 m at 20 km (conversion returns heights too low).

Geopotential height	Old default (63781370)	Correct (6378137)
10000 m	10001.568 m	10015.703 m
20000 m	20006.273 m	20062.912 m

Changes

• Defaults: radius=63781370(0.0) -> radius=6378137.0 (WGS-84 semi-major axis, matching the value used in environment.py geodesy and the existing docstrings) in all three helpers
• Docstring Default is 63781370 m -> Default is 6378137.0 m (WGS-84 semi-major axis)
• Docstring examples re-anchored to the correct values
• New regression test test_geopotential_height_to_geometric_height_default_radius pins the correct conversion at 10/20 km and checks explicit-radius override still works

Testing

• tests/unit/test_tools.py + doctests of rocketpy/tools.py: all pass
• tests/unit/environment/test_environment.py: 145 passed (explicit-radius call sites unchanged)
• black clean