[FEA] Support custom “Delta Degrees of Freedom” for `matx::var`

[ZJUGuoShuai]

Is your feature request related to a problem? Please describe.

I found that matx::var do not support the ddof argument like NumPy var. This parameter determines how to do the final division step.

ddof is 0 by default in NumPy, whereas in MatX it is currently fixed at ddof=1. Is it possible to support having the user pass in a custom ddof parameter? A similar feature is implemented by PyTorch, in their var function, where the equivalent argument is correction.

Describe the solution you'd like
I want a var function which looks like

void var(OutType dest, const InType &in, const int ddof = 1, int stream = 0)

[cliffburdick]

Thanks for the request. That seems easy enough and we will add this soon.