Call close() and join() on multiprocessing.Pool()

According to the documentation
https://coverage.readthedocs.io/en/latest/subprocess.html#using-multiprocessing
a `pool` needs to be closed and joined, even inside a `with` block.

According to
https://docs.python.org/3/library/multiprocessing.html#multiprocessing.pool.Pool
the `Pool` class calls `terminate()` whenever the context manager calls `pool.__exit()__`
This will terminate the worker processes immediately, and apparently `coverage` won't have
enough time to record its coverage measurements.

So this change is only needed to measure the coverage properly.

lobster/tool.py

@@ -254,5 +254,7 @@ def execute(self):
                 for new_ok, new_items in pool.imap(pfun, work_list, 4):
                     ok    &= new_ok
                     items += new_items
+                pool.close()
+                pool.join()
 
         return self.write_output(ok, options, items)