Experiment with ways to add synthetic content to Web servers

This is a child issue of #1.

HTTPT relies on its "fronting" Web server to not look suspicious so that censors cannot find and block these Web servers easily. In other words: the index page of the Web server shouldn't say "This is a hidden HTTPT proxy." If we bundle a Web server with an HTTPT server (rather than assume that the HTTPT operator already runs a Web server), we need a way to create "natural-looking" Web content. Section 3.2 of the HTTPT paper lists several options.

Another option would be to experiment with OpenAI's GPT-3 and see if we can use it to automatically generate, say, a blog. Let's use this issue to make progress.