FeedBunch can use a remote browser instead of a local headless browser

When fetching certain feeds a full browser is necessary (e.g. to pass certain anti-DDoS measures). Until now a local Chromium was started every time such a feed was fetched, with required chromium and its webdriver to be installed in the local machine. This can be problematic when running FeedBunch in a dockerized environment, because running a chromium browser process in a container is not trivial. With these changes a remote browser can be used instead. The HEADLESS_BROWSER_LOCATION environment variable controls whether a local (value "local") or remote (value "remote") browser is used. The default is to use a local browser. If a remote browser is used, HEADLESS_BROWSER_HOST env variable controls its host (defaults to 127.0.0.1) and the HEADLESS_BROWSER_PORT env variable controls the port where it's listening for Selenium connections (defaults to 4444). The easiest way to get a standalone chrome browser that listens for selenium connections is to run the selenium/standalone-chrome docker image.