Scrape GitHub data with fewer API requests

[choldgraf]

Currently we use up quite a lot of API requests every time that we scrape the GitHub data. This means that if we update the site too many times, we'll hit our limit and have to wait a day. We should find a way to scrape this data in a way that uses fewer API calls. For example, by storing the data in a persistent cloud bucket and simply updating that data rather than scraping from scratch each time.