docs(splat-native): sync module skeleton with §4.2/§4.3 (codex P2 on #213)

Follow-up to codex review on PR #213. The per-primitive specs at §4.2
(`batched_mahalanobis`) and §4.3 (`batched_opacity_blend`) were
correctly updated in #213, but the module skeleton at the top of the
same file (the `pub fn` declarations in the `src/simd_splat.rs` sketch
around line 50) still advertised the OLD no-scratch / no-ray-offsets
signatures.

Since this document is the implementation handoff for
`src/simd_splat.rs`, an implementer scanning the skeleton first would
recreate the exact APIs #213 was meant to eliminate, leaving the
zero-allocation contract (Mahalanobis) and per-ray segmentation
contract (opacity blend) unenforceable.

## Fix

Updates the two skeleton signatures to match §4.2 and §4.3 exactly,
with cross-references to the per-primitive sections:

- `batched_mahalanobis` — adds `cholesky_scratch: &mut [f32]` (length
  N × 6); inline comment cites §4.2 sizing guidance and the
  "function MUST NOT allocate" contract.
- `batched_opacity_blend` — adds `ray_offsets: &[u32]` (length
  n_rays + 1, CSR-style); `sorted_amplitudes` re-described as flat
  concatenation; `out_alpha` length now `n_rays` (not "per pixel");
  inline comment cites §4.3 segmentation contract.

The two implementations of the §4.2 / §4.3 detailed specs are
unchanged in this PR — this commit only syncs the up-front skeleton
so the two views of the API agree.

## Test plan

- [x] Skeleton + §4.2 + §4.3 now show identical parameter lists.
- [x] Comments on the new parameters cite the section that owns the
      detailed contract (so an implementer who reads the skeleton
      first gets pointed to the contract section).
- [ ] Codex re-review on this PR.

Author: claude

.claude/plans/splat-native-ultrasound-simd-substrate-v1.md

@@ -48,16 +48,18 @@ pub fn batched_cholesky_3x3(
 );
 
 pub fn batched_mahalanobis(
-    query_xyz: &[f32],        // M × 3 query points
-    mu_xyz: &[f32],           // N × 3 Gaussian centroids
-    sigma_packed: &[f32],     // N × 6 packed Σ
-    out_dist_sq: &mut [f32],  // M × N output (squared Mahalanobis)
+    query_xyz: &[f32],            // M × 3 query points
+    mu_xyz: &[f32],               // N × 3 Gaussian centroids
+    sigma_packed: &[f32],         // N × 6 packed Σ
+    cholesky_scratch: &mut [f32], // N × 6 — caller-provided packed-L scratch (24 MiB @ N=1M); function MUST NOT allocate (see §4.2)
+    out_dist_sq: &mut [f32],      // M × N output (squared Mahalanobis)
 );
 
 pub fn batched_opacity_blend(
-    sorted_amplitudes: &[f32], // N (front-to-back along view ray)
-    opacity_lut: &[u8; 256],   // amplitude → opacity LUT
-    out_alpha: &mut [u8],      // composited alpha per pixel
+    sorted_amplitudes: &[f32],    // flat; all rays' samples concatenated (front-to-back per ray)
+    ray_offsets: &[u32],          // length = n_rays + 1 (CSR-style); ray r's range is [ray_offsets[r]..ray_offsets[r+1]) (see §4.3)
+    opacity_lut: &[u8; 256],      // amplitude → opacity LUT
+    out_alpha: &mut [u8],         // length = n_rays — composited alpha per ray
 );
 
 pub fn batched_sh_eval_l3(