Change Record._endpoint_from_url() path parsing to ignore host:port

[markkuleinio]

This PR changes the logic in Record._endpoint_from_url(): with this change it does not matter what's in the host:port part of the URL as we always check the path of Api.base_url, and if it contains some extra leading directories, those extra directories will be removed from url before parsing the app and name.

There are two sides in this:

1. This enables the path parsing to work even if the host:port is for some reason different in the NetBox-returned URLs and in the pynetbox user-supplied Api.base_url.
2. This also masks out some reverse proxy configuration errors like in API .update() fails when API host is in the format "http://ipaddress". Endpoint API URL has missing characters #322 (but that was the original behaviour before Fixes #304: Change Record._endpoint_from_url() URL parsing to support other paths for NetBox than root #308 introduced in pynetbox 5.1.1).

[zachmoody]

👍 Thanks! Assuming we can skip normalization now.