verifying website or domain ownership issues HTTP GET without User-Agent header

Keybase should always specify a `User-Agent` header when making HTTP requests.

Keybase issued the following HTTP request to my web site when I attempted to verify ownership of my domain:

```
GET / HTTP/1.1
host: skepticism.us
Connection: close
```

The lack of a `User-Agent` header caused my firewall to blackhole the source IP address. Which caused the subsequent `GET /keybase.txt HTTP/1.1`, which did include a `User-Agent: keybase-proofs/2.0.48` header, to be rejected. I do this because a legitimate request should always include a user-agent while a lot of malware requests do omit it. Yes, it's a very simplistic rule but it almost never results in a false positive and does catch attacks my other rules might not catch.

Locate the code that performs the initial HTTP GET during website or domain ownership verification, then compare it with the subsequent request to /keybase.txt described in the issue. Done means both requests include a User-Agent header and verification works for sites that reject requests without one.

Written by the indexing model from the issue text.