Update naming conventions and config structure

[charles-turner-1]

Config.yaml currently contains the following:

catalog:
  endpoint: /intake/update
  items:
    - esm_datastore.search
    - DfFileCatalog.search
    - DfFileCatalog.__getitem__
payu: 
  endpoint: /payu/update
  items:
    - Experiment.run

which is of the more general form

service_1: 
  endpoint: endpoint_1
  items:
    - s1_li_1
    - s1_li_2
service_2:
  endpoint: endpoint_2
  items:
    - s2_li_1
    - s2_li_2
service_$N

This structure is necessary in order to respect the structures of the ApiHandler and TelemetryRegister classes. However, it enforces that we can only have a single endpoint per service. Conceptually, a service should really map to a Django app, and an endpoint to a model serializer, and so as things currently stand, we can only have one model per app, unless we resort to confusing tricks like

catalog:
  endpoint: /intake/update
  items:
    - esm_datastore.search
    - DfFileCatalog.search
    - DfFileCatalog.__getitem__
payu_run: 
  endpoint: /payu_run/update
  items:
    - Experiment.run
payu_restart: 
  endpoint: /payu_restart/update
  items:
    - Experiment.run

This needs fixing, so we have a schema more like

service_1: 
  endpoints:
    - endpoint_1:
      - s1_li_1
      - s1_li_2
    - endpoint_2:
      - s1_li_1
      - s1_li_2

...

eg

payu: 
  endpoints:
    - run:
      - Experiment.run
    - restart:
      - Experiment.restart
catalog:
  endpoints:
    - update:
      - esm_datastore.search
      - DfFileCatalog.search
      - DfFileCatalog.__getitem__
...

[charles-turner-1]

This can be further simplified by building the endpoints procedurally from the yaml structure: eg.

catalog:
  update:
    - esm_datastore.search
    - DfFileCatalog.search
    - DfFileCatalog.__getitem__
payu:
  run:
    - Experiment.run
  restart:
    - Experiment.restart

where calls to esm_datastore.search, DfFileCtalog.search, or DfFileCatalog.__getitem__ would send post requests to /catalog/update etc.