Feature tensor zeros

[byte-6174]

After an interesting deep dive into how torch initializes tensors with zeros(1), and several detours on how calloc and malloc actually behave (2) + why it is better to use calloc than malloc+memset (3), decided to use calloc for initializing tensor with 0s.

Also learned that Memory coming from the OS will be zeroed for security reasons.* ! (4)

---

(1) - Uses memset (https://github.com/pytorch/pytorch/blob/50595ecef4a4f9882a02539019b11a5e50295244/aten/src/ATen/native/Fill.cpp#L150)
(2) - https://stackoverflow.com/questions/1538420/difference-between-malloc-and-calloc
(3) - https://stackoverflow.com/questions/2688466/why-mallocmemset-is-slower-than-calloc?lq=1
(4) - https://stackoverflow.com/a/8029624/2697940