erp5_wendelin_data_lake_ingestion:  split files are no longer appended/processed and no data streams are removed anymore
- all ingestions and data streams corresponding to split parts are kept
- client will receive the list of all data streams and it will be in charge of merging the parts during the download
- validate chunk data streams only when full file was ingested
- only process split ingestions when full file was ingested
- calculate full split file size
- calculate hash and add state control during data stream validation
- stop invalidated ingestions

- do not return invalid data streams (filter by reference and state)
- groups split file datastreams by reference
- returns full file size and large hash

erp5_wendelin_data_lake_ui: fix UI queries to get documents and update renamed server scripts in urls

See merge request !44

categories. This is usually needed for Wendelin applications where for some
reasons data is stored in different Data Streams which must form a logical and sorted "link".
Add test.

See merge request nexedi/wendelin!42

See merge request !41

Allow owner of a Data Stream to edit it despite published. In normal ERP5 this is not normal but with Data Lake functionality validation is done at ebulk side (which also takes care of increasing automatically Data Set's version). This is quite simplified model of ebulk only validation of a Data Set but sufficient for most cases.

Rename validation_workflow (which duplicates erp5_core's one.) to data_validation_workflow which ash one new state only 'Published'
Cleanup worklists.

This reverts commit ab6250a1.

Validates state's meaning is that it has passed a validation by content - either ebulk client side or Wendelin side. It's not meant to be part of publication context.

Make it possible to configure if default (open) security model is to be installed for Wendelin's Data Lake. Default is not install.

Add new states for Data Stream and Data Set so we can have a more complex security models (based on DMS' publication_workflow).