Improve test coverage for dxtbx_machinery module

[coderabbitai]

Context

Test coverage is needed for the mdx2.dxtbx_machinery module to verify edge cases and improve code maintainability.

Specific Example

The calc_U_matrix_at_phi function has a potential edge case where the base index b from phi_to_base_fraction_index() may not be properly bounded by crystal.num_scan_points. This happens when crystal.num_scan_points != scan.get_num_images(), potentially causing an out-of-bounds access when calling crystal.get_U_at_scan_point(b).

Requested Action

Generate unit tests that:

• Cover the edge case mentioned above
• Test boundary conditions for scan-varying and non-scan-varying crystals
• Validate behavior when crystal.num_scan_points differs from scan.get_num_images()
• Improve overall test coverage for the dxtbx_machinery module

References

• PR: Spm/44 prepare new release #45
• Comment: Spm/44 prepare new release #45 (comment)
• Requested by: @spmeisburger
• Note: This issue replaces Improve test coverage for dxtbx_machinery module ando-lab/mdx2#10 (created in wrong repository)

[coderabbitai]

Additional edge case: calc_B_matrix_at_phi boundary handling

Similar to the concern raised for calc_U_matrix_at_phi, the calc_B_matrix_at_phi function has the same scan-point boundary issue.

Location: mdx2/dxtbx_machinery.py, lines ~441-459

Issue: When crystal.num_scan_points > 0, the function calls crystal.get_B_at_scan_point(b) where b comes from phi_to_base_fraction_index. If the scan and crystal scan-point counts differ, b can be out-of-range, potentially causing an index out-of-bounds error.

Current mitigation in PR #45: The code now includes boundary clamping logic similar to calc_U_matrix_at_phi:

• Clamps b2 = min(b + 1, imax - 1)
• Uses direct value (no interpolation) when at boundary

Testing needed:

• Unit tests for edge cases where scan-point indices are at boundaries
• Tests for mismatched scan/crystal scan-point counts
• Validation of interpolation behavior at endpoints

Related: PR #45 comment thread at #45 (comment)

cc: @spmeisburger