Expose the slice step that affine_of_package already computes (#177)

[gdevenyi]

Fixes #177.

affine_of_package builds the centre-to-centre slice step, uses it as the third column of the affine, and discards it. This exposes it:

@property
def slice_distance(self):
    """Centre-to-centre slice spacing of each slice package, in mm."""
    return [float(np.linalg.norm(self.affine_of_package(package)[:3, 2]))
            for package in range(len(self.slice_packages_index()))]

Shape and availability are as the issue proposed: one value per package, because packages can be spaced differently (spec 7.10), and available exactly where affine_of_package is — note that is affine_of_package, not affine, so a parametric map whose third axis is a frame group still reports its spacing.

SlicePackageSplitter._split_VisuCoreSlicePacksSliceDist had the same norm inline; it now asks for it.

Measured on our corpus

1631 vendor 2dseq reconstructions (PV 5.1, 6.0, 6.0.1, 7.0.0, 360.3.4–360.3.7), against np.linalg.norm(affine_of_package(i)[:3, 2]):

source	disagrees with the affine
VisuCoreSlicePacksSliceDist, falling back to VisuCoreFrameThickness	197 / 1599
resolution[2]	82 / 1599

slice_distance itself is finite and positive for all 1599, and unavailable for the 27 that carry no VisuCorePosition/VisuCoreOrientation. 0 bad values.

Both alternatives fail structurally rather than marginally:

                                   parameter      affine
T1_FLASH_3D_iso                         12.0       0.125    VisuCoreFrameThickness is the slab
UTE3D                                   25.0       0.195
aswendt_gfap_pt/GV_T3_1_1_baseline/3    25.0       0.391

T2star_map_MGE/pdata/2                   0.0       1.000    resolution[2], third axis not spatial
lego_phantom/3                     28.284272       1.000    resolution[2], 20*sqrt(2) across packages

Of the 82 resolution[2] disagreements, 13 are the 0.0 parametric-map case and 61 are the cross-package diagonal.

Notes for review

• Not added to COMPUTED_REPORT_PROPERTIES. affine is in bruker report output and this arguably belongs there too, but that changes the report schema and all four property references, so I left it as an API-only addition for you to decide on. Happy to add it.
• Tests are synthetic (test/synthetic.py), covering the gapped stack, the 3-D slab, the per-package list, the parametric map, and the no-geometry failure.
• Suite: 2114 passed, 12 skipped. The 5 test_affine_agrees_with_the_paravision_nifti_sform failures are the same ones master has.

🤖 Generated with Claude Code

https://claude.ai/code/session_01NuK1cZi8U54WXAdXMmGpzy

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.