"just in case" release for UTF-8 bug

Making 100% sure every SemTK change is in the RACK -dev docker release, because UTF-8 bug #320 is not reproduceable in dev environment but still happens in docker.

Fix ingestion lock-up on certain no-op rows

v2.3.0-20210322

Fix ingestion bug where it can lockup on  a poorly-placed row with no…

subProperties change and new display of invalid nodegroups

subProperties no longer infer Domain from their super-properties.

invalid nodegroups loaded into SPARQLgraph display a full list of errors and a visual display.

Start subset of services

ENV_OVERRIDE setting of ENABLED_SERVICES allows a subset of SemTK services to be started, leading to less resource-hogging docker containers, for one thing.

Other minor improvements

Minor bug fixes

Aligns with RACK release 5

CONSTRUCT with UNIONS and BINDINGS

More improvements and regression testing on CONSTRUCT queries

Improved ingest performance fuseki and blazegraph

Optimized the URILookup queries used during ingestion.
Different optimizations for fuseki vs blazegraph.
Massive performance improvements for these two triplestores ingesting with URI Lookup.

CONSTRUCT queries and bug fixes

Vastly improved CONSTRUCT query functionality, especially for Fuseki.
Support two properties with the same KeyName
Various small bug fixes

Date format and double URI lookup

Major fix: correct logic to recognize two identical URILookups in the same ingestion template and guarantee uniqueness.

Also added the default SADL datetime format.

SPARQL gen improvements

Handles ontologies where a class has two properties with same "keyname" but different full URI.
e.g. http://one/hasProp and http://two/hasProp

Fixed sparql generation of union queries where one union branch is a class with an incoming property.

Various other bug fixes.