Fix error codes in tests being emitted as IRIs

[rubensworks]

Currently, the expected error codes in NegativeEvaluationTests are defined using the expect context term.
However, this term is defined with "@type": "@id", which causes the error codes to become IRIs, which is incorrect, as the spec describes them to be plain strings.

This PR adds a new expectErrorCode context entry, which expands to the same predicate as expect, but as a plain string. All NegativeEvaluationTests have been updated to make use of expectErrorCode instead of expect.

[gkellogg]

Seems reasonable, but we’ll need to update the README and template.md accordingly, and perhaps tweak the test vocabulary definition. Also, please use the rake task to regenerate the HTML versions of the test manifests.

[rubensworks]

@gkellogg I've applied the changes you requested.

I did not change README.md, as it did not have any explicit mentions to expect, and instead refers to the vocab, where the distinction between expect and expectErrorCode is mentioned. (If you think this would be beneficial, I can add this into README.md as well)

I'm satisfied.

[gkellogg]

Thanks, Ruben. This is great. If running the tests is unclear, we'll get feedback.

Can you do the same for the framing tests?

[gkellogg]

I'll take care of the framing tests, as you don't work with that.