Implements node subsampling methods for multivariate inference on
network moments (rescaled motif counts), including: uniform node subsampling
to approximate the joint distribution of multiple network moments (Algorithm 1);
externally sparsified moments for density-matched comparisons (Algorithm 2);
and a two-sample test for unmatchable networks with unequal edge densities
via a split-and-sparsify subsampling procedure (Algorithm 3). Built-in support
for V-shape (2-star), triangle, and 3-star motifs, with a user-extensible
interface for arbitrary additional motifs. Parallel execution is supported
via 'doParallel' and 'foreach'. Based on Qi, Hua, Li and Zhou (2024)