How could we reduce CI timeouts with caching?

## Problem

In CI, we experience failures for PEXEX on a somewhat regular basis across ~80 test cases (for upstream and unicorn flavors). One cause of failures that we identified was with test environment setup as opposed to the test itself. Sometimes, installing k3d or running `npm install` will result in failures due to connectivity issues.

## Focus

We may be able to use caching or a new runner images with those tools already present to avoid potential timeouts in CI that force us to retry PEXEX tests.

## Goal

This spike should consider our options and enable us to choose a path forward to avoid k3d or npm install failures in CI.