Matrix Multiplication: Cmma Matrix Double Buffering

[nathanielsimard]

Blocked by #357

The goal is to be able to create an array of matrix fragments. To do that we'll need to refactor the cmma::Matrix type to contain the ident, matrix shape and matrix kind:

MatrixFrag<FragC, (f16, f16, f32), (16, 16, 16), Accumulator>

With that we'll be able to create an array like that:

let mut rhs_frags = Array<MatrixFrag<FragB, (f16, f16, f32), (16, 16, 16), RowMajor>>::new(2);