Use along-strike length when converting the FSP strike bearing

[lispandfound]

Fixes #91

Problem

FspSegment.to_plane converted the great-circle strike to an NZTM bearing but passed the segment's width as the shift distance:

# before
strike_nztm = coordinates.great_circle_bearing_to_nztm_bearing(
    self.top_centre, self.width, self.strike
)

The distance parameter is the distance shifted along the bearing being converted (qcore/coordinates.py:193-194; the implementation calls geo.ll_shift(x, y, distance, great_circle_bearing)). Since the bearing here is the strike, the distance should be the along-strike extent.

srf.py:608-612 performs the same conversion for a strike and correctly passes the along-strike segment_header["len"]. Within to_plane itself, self.width is legitimately the distance for the dip direction a few lines below, which is where the confusion came from.

Scale of the effect — small

Measured at origin (-43.5, 172.6) for a great-circle bearing of 45 degrees:

distance =  5.0 km -> 44.826196
distance = 20.0 km -> 44.826554
distance = 60.0 km -> 44.827250

About 0.001 degrees across a 5-60 km range: the NZTM convergence correction is dominated by origin latitude, not by distance. So this is a correctness and clarity fix, not one with visible consequences — flagged because the argument is plainly the wrong quantity and will mislead the next reader. I would not prioritise merging it.

Changes

• source_modelling/fsp.py — pass self.length.

No test added: the difference is far below any tolerance a meaningful assertion could use, so a test would either be vacuous or pin down floating-point noise. The srf.py precedent is the real evidence.

Verification

• pytest tests/test_fsp.py — 158 passed.
• ruff check, ruff format clean.

🤖 Generated with Claude Code