Peter van der Graaf has written about similar setups:
http://www.vdgraaf.info/google-analytics-without-javascript.html

You would need to filter on user agent however to get the browser requests and not the Desktop feed readers/bots.
What's more: Mozilla/Flock often prefetch feed urls in the autodiscovery info, without the user actually subscribing. So you might want to measure subscriptions by measuring how many people follow the "path" (1 - request to feed url) (2 - request to image included in the html)

You are basically lying to the browser about the content type. You are telling the browser it should expect application/rss+xml at the alternative URL, but it's plain HTML instead.

I can actually lead to the inability to subscribe to your feed. For example try subscribing to your feed from Safari, it won't work and Safari will show an error message instead.. Also external feed readers usually register themself at the operating system for the application/rss+xml type and fetch the alternative URL themselves. If they don't understand HTML and Javascript - and I doubt any of them do - those feedreaders will never reach your actual feed URL.