When the user navigates to a page showing site exceptions, for example chrome://md-settings/content/flash, there are three <site-list> instances on the page, one for each of allow, block and session-only.

I discovered inneficiencies regarding how/when those three lists get populated:

1) When a site exception is added/removed, all three <site-list> instances
   re-render themselves, even if nothing changed in that list.
2) All three instances query the browser (via a browserProxy) on their own and
   get back the exact same info (a list of all site exceptions returned 3 times
   instead of just once). Then each list iterates over that list and filters out
   the exceptions that belong to other lists (the "allow" list will filter out
   "block" and "session-only" etc).

For 1: It would be more efficient if only the things that changed re-render themselves.

For 2: The three lists already have a common parent element <category-setting-exceptions>. It would make more sense if the parent was making a single call to the browserProxy, split the returned exception list to three lists and pass those to its children respectively.