Performance of vocabulary validation

[ledsoft]

When validating a moderately-sized vocabulary (ca 400 terms) with flat structure validation often times out on the proxy after 60 seconds. This is unacceptable.

[ledsoft]

Update: the validated vocabulary imports another vocabulary containing ca 660 terms, so the total number of terms validated is around 1000.

[ledsoft]

TTL export of the relevant vocabularies (which is what the Validator uses - it exports the vocabulary contexts into a byte stream and imports it to a Jena in-memory model) has approximately 1.4MB.

[ledsoft]

Did a quick comparison with termit-dev, which does not run in Docker and is thus able to utilize all the CPUs on the kbss server. It was able to validate a vocabulary with 1800 terms in 22 seconds (note that GraphDB it uses is on a different server). Will need to try testing it locally in Docker so that CPU availability can be tuned for the test.