Clarify not found error handling when dataset, etc not found

Currently, getting a dataset with a given PID returns 200 and an empty body when the dataset does not exist. The same applies to the origdatablocks endpoint. (I didn't test other endpoints.)

In my mind, those should return 404 instead as the requested dataset, etc., does not exist.

Alternatively, we could interpret a 200 response with an empty body as 'object not found'. But this requires that no endpoint (with GET request) returns an empty response that means something other than 'not found'. If we go this route, this behaviour should be documented.

All endpoints should behave the same way, regardless of which solution we go with.

To reproduce, e.g., use this with SciCat deployed from the stable image:

import requests
from urllib.parse import quote_plus

r = requests.get(
    url="http://localhost:3000/api/v3/Datasets/" + quote_plus("abcd/123"),
)
print(r)
print(f'"{r.text}"')

prints

<Response [200]>
""

Start with the GET /api/v3/Datasets/{pid} entry point and compare it with the origdatablocks endpoint using the reproduction request in the issue. Review the other GET endpoints for consistent empty-response behavior, then verify that all endpoints follow the chosen not-found convention and add or update coverage where the project supports it.