Pin Apple Silicon half conversion fixes

[gripeyes]

Compatibility:

patch

Issues/Tickets:

Native AOVs rendered through Houdini on Apple Silicon could contain corrupted or zero-filled values before these fixes.

OpenMoonRay/moonray#36
OpenMoonRay/scene_rdl2#25

Release notes comment:

Updated the MoonRay and scene_rdl2 revisions to include Apple Silicon scalar half-float conversion fixes.

Comments for the reviewer:

This PR updates only two submodule pointers:

• moonray/moonray to Fix ARM half conversion in RenderOutputWriter moonray#36
• moonray/scene_rdl2 to Fix ARM half conversion in PackTiles scene_rdl2#25

Both source changes replace unsafe ARM/NEON vector load/store operations on scalar addresses with scalar __fp16 conversion and std::memcpy.

The related hdMoonray render-buffer allocation change from the original investigation is intentionally excluded.

This is a stacked PR and depends on both source PRs. If either source PR is squash-merged, this branch will need to be updated to its final upstream commit.

Look or scene setup change:

No scene setup changes are required. Apple Silicon outputs affected by incorrect half-float conversion may now contain the intended values.

Special notes for production:

The source changes affect ARM/NEON builds. Existing non-ARM conversion paths are unchanged.

Attention/Reviewers:

AI Assisted Development:

Assisted-by: OpenAI Codex / GPT-5.6

Checklist:

• Documentation has been updated.
• Includes new unit tests.
• Includes new RATS tests.

No documentation, unit-test, or RATS changes are included in this submodule-pin update.