Stop using TaliskerRequestsTransport

At least for now stop using the TaliskerRequestsTransport.

In practice, for flask apps it was not used in the apps/workers because
FlaskSentry was explicitly setting a None 'transport' in the client
kwargs.

This change makes sure that the same default raven (threaded) transport
is used by logging in the main process. This will mitigate the deadlocks
described in
prometheus/client_python#1076 (comment)
because there will be no prometheus instrumentation for sentry requests.

talisker/sentry.py

@@ -317,9 +317,6 @@ def ensure_talisker_config(kwargs):
     processors = set(kwargs.get('processors') or [])
     kwargs['processors'] = list(default_processors | processors)
 
-    if 'transport' not in kwargs:
-        kwargs['transport'] = TaliskerRequestsTransport
-
     # note: style clash - sentry client api is 'sanitize_keys'
     sanitise_keys = kwargs.get('sanitize_keys', [])
     if sanitise_keys is None:  # flask integration explicitly sets None