Don't treat some HTTP status error codes as not errors (#78)

Previously 401, 403, 503, 504, and 999 HTTP status codes were
treated as "not an error" and therefore not reported. At the time
this made sense as there was no easy way to say "do not treat 504
codes as an error, I will assume it is transient". However, with a
recent change you can do:

```yml
raise_error_excludes:
        504: ['*']
```

And treat all 504s as not an error.

This commit makes all "special HTTP status codes" an error and allows
for users to configure their `raise_error_excludes` if they want to
get the old behavior back.

I found this functionality when trying to figure out why some links
that we had that there bad weren't being reported by htmlproofer. It
turned out that the broken links were coming back with "probably good"
error codes and thus we never saw them.

Closes #77

htmlproofer/plugin.py

@@ -285,17 +285,10 @@ def contains_anchor(markdown: str, anchor: str) -> bool:
     def bad_url(url_status: int) -> bool:
         if url_status == -1:
             return True
-        elif url_status == 401 or url_status == 403:
-            return False
-        elif url_status in (503, 504):
-            # Usually transient
-            return False
-        elif url_status == 999:
-            # Returned by some websites (e.g. LinkedIn) that think you're crawling them.
-            return False
         elif url_status >= 400:
             return True
-        return False
+        else:
+            return False
 
     @staticmethod
     def is_error(config: Config, url: str, url_status: int) -> bool: