“FastDEC” operators for 3D

Currently, we have two implementations of most DEC operators. One is a reference implementation, and the other a faster version that operates over global chains rather than computing the value at some particular simplex. These fast operators may be kernel operations, matrix vector operations, or some combination. Further, the pre-allocations (or caches) of these operators may also be optimized, such as in the case of constructing the exterior derivative matrix.

However, these fast implementations mostly concern the 2D DEC. As we move towards more 3D simulations, we should implement fast versions of these operations as well.

So, we should focus on fast implementations of the elementary derivative and Hodge star operations. Note that PR #165 introduces fast matrix versions of wedge product operations for the 3D DEC.