FPM-1338 shared processing configs

[richjam]

https://jira.ceh.ac.uk/browse/FPM-1338

Overview

A data processing configuration can apply to more than one dataset, e.g. fdri-se-fofad-01-pr-battv_1min-load_raw is a step in both the battv_15min and battv_30min plans, so the metadata API returns it with two appliesToDataset values.

Previously, we assumed a 1-to-1 relationship so did applies_to_dataset[0], so the config was set against whichever dataset the API happened to list first and dropped from the others. One of the datasets would fail with KeyError: 'http://fdri.ceh.ac.uk/id/data-processing-configuration/fdri-se-fofad-01-pr-battv_1min-load_raw'

Main changes

• map_processing_config_item now takes the specific AppliesToDataset entry it is mapping for, and reads ts_id
  and site_id from that entry rather than from applies_to_dataset[0].
• _build_processing_configs now takes the dataset IDs the configs were requested for. Each config is mapped once per
  requested dataset it applies to, so which container a config lands on no longer depends on the order the API lists
  those datasets in