Make the whole fields extraction + weights computation part of the catalogue search API

[minottic]

I think to ease every RIs deployment and adoption we should create one new endpoint in every data catalogue search API (e.g. in the SciCat search-API), which takes a bearer token and triggers the whole flow (let's say /compute-weights).
Namely:

1. using a json file provided to the search-API we define the collection+fields we want to take from the (scicat) DB
2. the /compute-weights does all the data preprocessing (i.e. for every item in each collection extract the relevant fields and compose the body of the subsequent post to the scoring service)
3. after 2, the /compute-weights posts the data to the compute service using "pss_items_url"
4. after 3, the /compute-weights posts the "pss_compute_url"

2,3 and 4 could be implemented in disjoint endpoints triggered sequentially if preferred.

It's left to every facility to schedule the /compute-weights computation and do the authz/auth to scicat.