Fix bug preventing API calls requiring a file ID

For example, an API call to files.info takes a file ID argument named
"file", which was stripped out by this call. Currently, there is only
one request type that accepts file data (files.upload). Every other use
of 'file' is an ID that aught to be contained in the request.

Author: Armin Samii

slackclient/_slackrequest.py

@@ -22,7 +22,12 @@ def do(token, request="?", post_data=None, domain="slack.com"):
         post_data = post_data or {}
 
         # Pull file out so it isn't JSON encoded like normal fields.
-        files = {'file': post_data.pop('file')} if 'file' in post_data else None
+        # Only do this for requests that are UPLOADING files; downloading files
+        # use the 'file' argument to point to a File ID.
+        upload_requests = ['files.upload']
+        files = None
+        if request in upload_requests:
+            files = {'file': post_data.pop('file')} if 'file' in post_data else None
 
         for k, v in six.iteritems(post_data):
             if not isinstance(v, six.string_types):