Shut down leaked executors in test_interpreter_pool

[StanFromIreland]

Several tests created an InterpreterPoolExecutor and never shut it down, so their worker threads were only cleaned up whenever the garbage collector happened to destroy the abandoned executor. On the slow buildbot that took longer than the 1-second limit threading_cleanup() allows in tearDown, for example in this run:

test_submit_func_globals (test.test_concurrent_futures.test_interpreter_pool.InterpreterPoolExecutorTest.test_submit_func_globals) ... Warning -- threading_cleanup() failed to clean up threads in 1.0 seconds
Warning --   before: thread count=0, dangling=1
Warning --   after: thread count=1, dangling=2
Warning -- Dangling thread: <Thread(InterpreterPoolExecutor-57_0, stopped 273622737248)>
Warning -- Dangling thread: <_MainThread(MainThread, started 273774698528)>
test_submit_local_instance (test.test_concurrent_futures.test_interpreter_pool.InterpreterPoolExecutorTest.test_submit_local_instance) ... Warning -- threading_cleanup() failed to clean up threads in 1.0 seconds
Warning --   before: thread count=0, dangling=1
Warning --   after: thread count=1, dangling=2
Warning -- Dangling thread: <Thread(InterpreterPoolExecutor-62_0, stopped 273622737248)>
Warning -- Dangling thread: <_MainThread(MainThread, started 273774698528)>
Re-running test.test_concurrent_futures.test_interpreter_pool in verbose mode