Sepiida

----
Fix bug in prometheus support if the before_request fails to run. This will happen if some middleware creates a response before our prometheus middleware runs

----
Add support for Prometheus metrics. You can subscribe to the metrics with

sepiida.prometheus.add_handler(app)

After that you can access the metrics at /metrics/

For now we only capture a few metrics - total request time, number of requests made with user sessions. Eventually we'll grow it

----
Removed the requirement of putting SEPIIDA_PAO_ROOT in your flask app config to use the groups module. It will now correctly just pull from your USER_SERVICE environment variable. The same is true of SEPIIDA_JWT_KEY

----
Remove the sepiida.endpoints.add_resource function. Replace it directly with Endpoint.add_resource function

You can convert old projects to the new pattern by taking any call of the form

sepiida.endpoints.add_resource(app, SomeEndpoint, 'foobar')

and converting to

SomeEndpoint.add_resource(app, 'foobar')

The logic is exactly the same. This just allows for subclasses to override the default behavior


I also changed json_client's response to have a json() method, not a json property. Yeah, lots of find-and-replace for that one. You're welcome. This makes the congnitive friction between this and aiohttp/requests lower

----
Fix the pao_group.uri fixture that got broken when I added a slash to the USER_SERVICE environment variable

----
Fix a few more poorly built URLs