This allows to check if a field should be stored as a custom field or an
extra

…ext-dcat into 56-add-schema-file-dcat-ap-2.1

Values for dct:publisher are obtained from the first one of these that
are present:

1. A scheming `publisher` field (assuming the `euro_dcat_ap_scheming`
   profile is loaded)
2. The legacy `publisher_*` extras
3. The dataset's own organization

For the last case, a sample schema for organizations has been added that
implements all the publisher properties supported by the processors.

New repeating subfield, supporting all properties for the location
class: uri, text, geom, bbox and centroid.
Used spatial_coverage as name to not interfere with the `spatial` field
expected by ckanext-scheming, in a future commit we will extract the
relevant value to index it as a geometry.

The previous field names based on indexes didn't allow to retrieve
results easily. We are now flattening all values for the same subfield
to at least get a text hit.

See #281 (comment)

If the `spatial_coverage` field is present, store the first geometry
found so ckanext-spatial can pick it up for spatial indexing.

Added indexing tests

At least the ones supported by the current processors.

TODO:
* spatial_resolution in meters: needs a new multiple_text_decimal validator
* hvd_category: will be done as part of the wider HVD work

This required a new scheming_multiple_number validator, adapted from
scheming_multiple_text

Support at the validator level for year, year-month, date and datetime
values, which are correctly typed in the RDF serialization.
At the UI level a date input is used by default as it was difficult to
provide one that supported all inputs.

Mostly taken from the DCAT-AP 2.1 spec doc, adapted for CKAN

Co-authored-by: Ian Ward <ian@excess.org>

As this is a `text` field that allows free text search

Scheming adds a dict with empty keys when empty
repeating subfields are submitted from the form. Check that there's an
actual value before creating the triples when serializing