Fix terminfo tests for API classification optimization

[johnbradley]

When using the v2-beta2 API changes to /term endpoint cause a test to fail.
The following line started failing:

rphenoscape/tests/testthat/test-pk.R

Line 208 in 531601b

testthat::expect_false("classification" %in% names(obj))

The above line tests that as.terminfo() doesn't fetch classification data when not using withClassification.
In the v2-beta2 release an optimization has been made to the underlying /term API endpoint to always return classification data.

Another aspect of the changed /term API endpoint is when a term is not found the classification structure is now filled in with empty lists. Example:

> obj <- as.terminfo("foo")
Warning message:
Failed to retrieve term for IRI foo 
> str(obj$classification)
List of 3
 $ subClassOf  : Named list()
 $ superClassOf: Named list()
 $ equivalentTo: Named list()

This causes another failure in the same test where the code expects obj$classification to be NULL for a bogus term:

rphenoscape/tests/testthat/test-pk.R

Lines 222 to 226 in 531601b

	testthat::expect_warning(obj <- as.terminfo("foo", withClassification = TRUE))
	testthat::expect_is(obj, "terminfo")
	testthat::expect_true(is.terminfo(obj))
	testthat::expect_false(is_valid_terminfo(obj))
	testthat::expect_null(obj$classification)

See phenoscape/phenoscape-kb-services#413 for details on the /term endpoint changes.

This is part of #235