disable tldextract caching

Fetching updated TLD lists is already disabled, so the default TLD list
that is included with the library will always be used and there is
nothing to cache. Explicitly disabling the cache prevents possible
occurences of john-kurkowski/tldextract#254.

pypac/wpad.py

@@ -9,7 +9,7 @@
 logger = logging.getLogger(__name__)
 
 # First use of tldextract goes online to update its Public Suffix List cache. Stop it.
-no_fetch_extract = tldextract.TLDExtract(suffix_list_urls=None)
+no_fetch_extract = tldextract.TLDExtract(cache_dir=None, suffix_list_urls=None)
 
 
 def proxy_urls_from_dns(local_hostname=None):