Fix request cache when filtering sensitive params.

[thhermansen]

This fixes an issue where filtering the URL actually mutated the given URL string. That string was also used as a cache key, but it didn't hit the cache as the string was mutated.

Fixes #32

PS. There are some a lot of rubocop related changes in this pull request as well as the fix. Sorry for that, but I didn't spend time keeping the rubocop commits separate from the fix.

[thhermansen]

This pull request filters the URL logged, but it uses the URL (with sensitive URL params) as cache key.

Maybe a more correct fix is to only use URL with sensitive data when doing a request to Google's API, but as cache key and when logging we'll use the filtered one.

What do you think, @aceunreal?

[thhermansen]

I think I'll push the code in the direction where we have a sensitive_url and a filtered_url from the UrlBuilder class., thus remove the responsibility of filtering the URL from the log subscriber.

Maybe we we could hash the sensitive_url when we add it to the cache as well, instead of use the filtered_url as hash key, as people may want to filter params which should be included as cache key.