Add a new 'wait' flag to the rescan command. This allows the caller

to request that the rescan command blocks until the rescan is
complete.

The reason is that if you run the tests on a Linux tmpfs they would
with high probability hang, due to the rescan completing before the
rescan_monitor had started up.

The flags is available in the Python interface but not the C interface
or the command-line client.  This could easily be fixed if there is
demand.

There's also a 'fresh' flag, to demand that the rescan start after the
receipt of the command (i.e. to guarantee your new tracks make it in)
but I disabled that due to the inconvenience of testing it.  However
the code is still there if anyone feels like writing tests.