Third person view treats cactus as a full cube

The bug

As far as collision, block targeting and the block's model are concerned, cactus is one less than a full block in each horizontal direction. Third person view, however, seems to treat the block as if it were 16x16x16.

The bug

1. Place two cacti such that their corners touch
2. Look through the corners of these cacti
3. Target blocks through the corners of these cacti
4. Attempt to push the third person camera between these two cacti

Expected results

Blocks would be visible and targetable through the gap in the cacti, and the third person camera would also be able to pass through this gap.

Actual results

Blocks are indeed visible and targetable through this gap, however the third person camera acts as though this gap does not exist.