[MAX-MAT-001] Strip buggy specular_rotation=0.25 from MaterialX export

[doug-smith-miris]

Summary

3ds Max's MtlxIOUtil.ExportMtlxString MaxScript bridge writes specular_rotation = 0.25 on every ND_standard_surface_surfaceshader, regardless of source PhysicalMaterial anisotropy. The MaterialX standard_surface nodedef default is 0.0, and specular_rotation is multiplied by specular_anisotropy in the BSDF — so when anisotropy is 0 (the universal default case), the value has zero observable effect, but it silently corrupts downstream pipelines that read MaterialX values directly or layer-override anisotropy later.

This PR adds a post-parse normalization step in MtlxShaderWriter::Write() that drops specular_rotation from standard_surface nodes when:

• the value is statically 0.25 (the bug pattern), AND
• specular_anisotropy is absent OR statically zero.

Connected anisotropy, any non-0.25 rotation, or a connected rotation input are all left untouched, so the fix is surgical and does not touch genuinely-authored anisotropic materials.

Diagnostic finding

• Fingerprint: MAX-MAT-001-specular-rotation-default-0.25
• Scope: 6/6 materials in samples/complex_export.usda affected
• Severity: medium — semantic correctness; silent regression risk for downstream anisotropy edits

What's changed

• src/translators/MtlxShaderWriter.cpp — adds _NormalizeStandardSurfaceSpecularRotation, called immediately after readFromXmlString.
• src/Tests/Integration/mtlxShaderWriter_test.ms — adds test_export_physical_material_strips_buggy_specular_rotation, a regression test that creates a default PhysicalMaterial, exports it via MaterialX, and asserts specular_rotation is either absent or 0.0.
• doc/translation-mapping.md — new tracking doc for 3ds Max -> MaterialX property translation status. MAX-MAT-001 is the first row.
• doc/changelog.md — Unreleased entry.

Validation

3ds Max is Windows-only and this repo builds via Visual Studio (configure-vsdevcmd.bat). I cannot build or run the plugin on macOS, so I validated by mirroring the C++ logic in Python at the USD layer (normalize_specular_rotation.py in the arch-build dir):

• Positive corpus — running the normalizer on samples/complex_export.usda strips exactly the 6 known-bogus specular_rotation = 0.25 lines and changes nothing else (USDA diff = 6 deletions, 0 additions, 0 modifications).
• Visual no-op — Karma renders of the pre-fix and post-fix USDs are byte-identical (same SHA-256: eaf3832840…b8351). This is the expected outcome: specular_rotation * specular_anisotropy = 0 when anisotropy is 0, so the lighting integrator never sees the value.
• Negative test — a synthetic fixture injects specular_anisotropy = 0.4 onto one shader. The normalizer correctly preserves specular_rotation = 0.25 on that shader while still stripping it from the other 5. Confirms the fix is surgical.

Build / run status

• Build (Visual Studio, Windows): deferred — requires a Windows MSBuild environment with the 3ds Max SDK; not available on this Mac.
• MaxScript regression test: deferred — requires 3ds Max (Windows-only).
• Logic equivalence: verified via Python validator against the captured corpus.

Status

Opening for review. The C++ change is straightforward and has clear semantics, but a Windows build/test pass is needed before this is merge-ready. The visual evidence and Python-validated post-fix USD strongly suggest the change is correct.