[ENH] More robust octree testing

Author: Leguark

docs/developers_notes/dev_log/log_2024-06.md

@@ -12,6 +12,37 @@
 
 - Create_grad_kernel is call 27 times
 
+### Wholes issue
+- Spremberg with 1.5 std of extra refinement:
+      Number of voxels marked by stats: 424 of torch.Size([512]).
+      Number of voxels marked total by corners: 281
+      Dense Grid would be 512 voxels
+      Number of voxels marked by stats: 2535 of torch.Size([3568]).
+      Number of voxels marked total by corners: 1492
+      Dense Grid would be 4096 voxels
+      Number of voxels marked by stats: 12148 of torch.Size([21152]).
+      Number of voxels marked total by corners: 7614
+      Dense Grid would be 32768 voxels
+      Number of voxels marked by stats: 69619 of torch.Size([101432]).
+      Number of voxels marked total by corners: 37423
+      Dense Grid would be 262144 voxels
+      Chunking done: 12 chunks
+- Spremberg with 1
+        Number of voxels marked by stats: 408 of torch.Size([512]).
+        Number of voxels marked total by corners: 281
+        Dense Grid would be 512 voxels
+        Number of voxels marked by stats: 1953 of torch.Size([3504]).
+        Number of voxels marked total by corners: 1492
+        Dense Grid would be 4096 voxels
+        Number of voxels marked by stats: 8632 of torch.Size([16856]).
+        Number of voxels marked total by corners: 7612
+        Dense Grid would be 32768 voxels
+        Chunking done: 13 chunks
+        Number of voxels marked by stats: 47653 of torch.Size([79080]).
+        Number of voxels marked total by corners: 37408
+        Dense Grid would be 262144 voxels
+        Chunking done: 10 chunks
+
 ### Other:
 {'_cached_pivot': [200, 200, 0],
 '_is_default_transform': False,