Fix app freeze on engine stop when rig disconnected

[tekstrand]

What

1. Moves rig teardown onto the TX handler
2. Removes releaseRigPttForShutdown and the unbounded CountDownLatch.await() the main thread sat in, along with the ad-hoc Thread {} that disconnected the network client.
3. Moves the transport construction out of onCreate into createRigTransports()

Why

Stopping the engine with the rig gone froze the app. The release waited on a latch with no timeout, and close() then ran in line behind it. Every HamlibRigControl method shares one monitor, so a wedged CAT command held both for hamlib's own timeout, and the USB bridge has the same shape where close takes a write lock a stuck read still holds. On my tablet with the radio pulled mid-transmission that was nine seconds of dead main thread, which is long enough for Android to call the app unresponsive.

Test

Need a rigctld to accepts the connection and never answer

For simulation without a real rig...

1. On your machine: python3 -c "import socket;s=socket.socket();s.setsockopt(1,2,1);s.bind(('0.0.0.0',4532));s.listen(1);c,_=s.accept();input()"
2. Settings, rig control on, type Network, host your machine's IP, port 4532.
3. Start, wait for Connected to rigctld in logcat, then Stop.
4. The UI should go back to Off right away, and Rig control torn down should turn up in logcat about five seconds later on the TX handler thread rather than the main one.
5. Start again. Rig control should reconnect.

On the emulator, before and after:

master
18:43:59.103  5745 5745  Stopping engine
18:44:04.158  5745 5745  RigCtlClient: NetworkOnMainThreadException
18:44:04.164  5745 5745  Engine stopped                          <- main thread, +5.06s

this branch
18:45:08.474  5880 5880  Stopping engine
18:45:08.537  5880 5880  Engine stopped                          <- main thread, +0.06s
18:45:13.542  5880 5913  Unable to confirm PTT release during shutdown
18:45:13.542  5880 5913  Rig control torn down                   <- TX handler, +5.07s