Client superfluously adding a trailing forward slash when provided with an empty path

https://www.rfc-editor.org/rfc/rfc3986#section-3
and
https://www.rfc-editor.org/rfc/rfc3986#section-3.3

Are http://base and http://base/ two distinct URLs? If so, so why does this is behaviour exist?

```
>>> from httpx import Client
>>> Client().build_request("GET",  "")
<Request('GET', '/')>
>>> Client().build_request("GET",  "/")
<Request('GET', '/')>
>>> 
```
