`Rack::Request#url` includes the port of the original request (despite
accurate scheme updates). This becomes an issue when LB terminates SSL:

- client request is made to https://example.com:443,
- LB request is http://example.com:80
- build_api_url returns correct scheme but wrong port:
  https://example:80

Since we only need the request url from the env, this changeset manually
fetches the correct scheme (Rack accounts for FWD headers already), the
host (w/out port as opposed to #host_with_port), and the fullpath
(pathname + query params)

Another benefit of this changeset is that it no longer mutates the
original env object (assigning to `new_env` is by reference).

Caveat: im dropping @options[:protocol] bc there's x-fwd-proto should
handle this already.

- Reinstating custom support for CF-VISITOR CloudFlare header +
  options[:protocol]

- Fixing failing tests
  Instead of passing in `X-Forwarded-Proto` to the Rack env, we can now
  safely rely on `HTTP_` env variables which are automatically set. Per
  https://github.com/rack/rack/blob/master/SPEC.rdoc#the-environment-:

      HTTP_ Variables:
      Variables corresponding to the client-supplied HTTP request headers
      (i.e., variables whose names begin with HTTP_). The presence or absence
      of these variables should correspond with the presence or absence of the
      appropriate HTTP header in the request. See RFC3875 section 4.1.18 for
      specific behavior.