Why doesn't TestPatternProxy work when "ran manually"?

Doing a full test suite run (UnitTest.runAll) on Windows seems hopeless as it dies some file access issues in TestLPCTestUtils: test_compareData_noDiff_oneEntry. So, for my purposes I’ve grepped the tests that actually have a chance of failing what I’m changing, namely pattern classes.

Thankfully I can run TestPattern.new.run and this does bunch of tests and all pass, but when I try TestPatternProxy.new.run (which might also be relevant) it passes immediately seemingly without running any tests, although clearly it contains some. (Annoyingly, there’s nothing like test counter printed, so I can’t tell how many it’s supposed to run.) Any ideas why TestPatternProxy.new.run basically does nothing visible?

-> a TestPattern
PASS: a TestPattern: test_pattern_zero_length - a Pfuncn : a pattern of length zero should return nothing but pass control (returned 1)

<< snip many more >>

PASS: a TestPattern: test_Psync_endsInTime - Pfindur  with maxdur = quant should end no sooner than after dur
PASS: a TestPattern: test_Psync_fillsRemainingTimeWithSilence - Psync with maxdur = quant should end no earlier than after dur

UNIT TEST.............
There were no failures
RUNNING UNIT TEST a TestPatternProxy
-> a TestPatternProxy

UNIT TEST.............
There were no failures