`clique_percolation` takes enormous memory and never finishes on a large graph

Description of bug
Hello, I was using the clique_percolation function to detect communities. It works for some of my networks, but for some other a bit more complicated ones, e.g. with 25376 connections between 2646 nodes, I found that the computation is using enormous amount of memory and simply won't finish. May I ask if this could possibly be a bug? Thank you for helping.

How to reproduce
Here I upload a network of mine in .jld2 format. You can simply unzip and read it in as
netw=load_object("network.local.jld2")
and transform it to graph using g=graph(netw). The computation I attempted is clique_percolation(g, k=3)

Expected behavior
Output results within a few hours or even minutes.

Actual behavior
Taking large memory (100+G) and never finishing.

Code demonstrating bug

netw=load_object("network.local.jld2")
g=graph(netw)
clique_percolation(g, k=3)

Additional context
Attachment:
network.local.jld2.gz

Start at the clique_percolation entry point and reproduce clique_percolation(g, k=3) with the attached network.local.jld2 data. The payload names no source files or tests, so first measure memory and runtime for this case; done means the computation completes without exhausting memory within the reported minutes-to-hours expectation.