in commit f25faf58a3ed81 the test runner was converted to a new API
which, instead of running single-step concurrently HDL-Simulator instruction 1,
HDL-Simulator instruction 2, .... it is now HDL instruction 1, HDL instruction 2
... followed separately by Simulator instruction 1, Simulator instruction 2 ...

HDL in SVP64 is not correctly detecting a terminating condition, instead
entering a permanent loop.  test_runner.py:

            terminated = yield self.issuer.dbg.terminated_o

this is never being set (in svp64 mode) and therefore the HDL simulation
never terminates.