Handle empty entries in the stratigraphy list

[perolavsvendsen]

If user puts empty entries in the stratigraphy in global_variables, e.g.:

MyFormation:
  stratigraphic: True
  name: My Fm. Top
  alias:
    - MyFormation
    - OtherAlias
    -                   # <-- PROBLEM!
  stratigraphic_alias:
    - Some GP. Top
    - SomeOther GP. Base   
    -                   # <-- PROBLEM!

---> This will currently export with null in outgoing metadata. In validation, it is expected to be string hence not validated. Tricky to debug. fmu-dataio should catch this and either warn or ignore or both.

Originally posted by @perolavsvendsen in #373 (comment)

Suggested solution:

• fmu-dataio gives warning when empty entries are detected, then ignores them

[perolavsvendsen]

Partial (?) solution in #386. However, does not ignore them - just warns.

A reflection: The whole validation of the config should probably be improved. This seems like a task for jsonschema or Pydantic or similar, and do proper validation of the config.

I tried jsonschema, but abandoned it for two reason:

• The nature of the "stratigraphy" block, with the custom keys, is not ideal for jsonschema
• The error messages coming out of json schema are too sparse to be informative to end user. E.g. if an empty entry is detected, it will not validate with a message that looks something like: "null is not a string".

Instead of validating the whole config in one go, we can create schemas for each entry, and check them one-by-one. That will enable us to add more meaningful information to the error messages.

[mferrera]

Closed in lieu of having completed the move to Pydantic which does this sort of validation.