bluetooth: Add FakeBluetooth interface

device/bluetooth/README.md

Index: device/bluetooth/README.md
diff --git a/device/bluetooth/README.md b/device/bluetooth/README.md
index f3c0ed8939fe811a3fa248b596cae8c584f85e38..991de6f0603c28544c576824accea7865bb9ad93 100644
--- a/device/bluetooth/README.md
+++ b/device/bluetooth/README.md
@@ -49,6 +49,13 @@ Known future work is tracked in the
 Implementation of the Bluetooth component is tested via unittests. Client code
 uses Mock Bluetooth objects:
 
+### Mojo Testing Interface
+A Mojo Testing Interface is defined at
+`src/device/bluetooth/public/interfaces/test/fake_bluetooth.mojom`.
+Clients can use this interface to control a fake implementation of the non-mojo
+C++ Bluetooth interface. The implementation of this test interface lives at
+`src/device/bluetooth/test/`. *This interface will be removed once a Bluetooth

[scheib, 2017/04/10 21:52:31]

'will be removed' is hard to say... ;) 'should' or 'may' is better. Perhaps
explain why: 

"""
This interface may be removed when a Bluetooth Mojo Service is introduced.
Testing code would implement the service as needed for tests.
"""

However, I also wonder how likely this is. This interface is a reusable test
interface that is likely to be pretty convenient to use. If multiple clients
need to test, it would save them effort vs each one creating a fake mojo service
impl, right?

[ortuno, 2017/04/12 05:49:06]

Done. I added that we'll remove it if Web Bluetooth is the only client.
I think it will depend on the approach we take when we add the Bluetooth
service. But based on the push back against refactoring clients to use the mojo
service directly and insistence on just using the current interface (T_T) my
guess is that we won't need this.

+Mojo Service is introduced.*
 
 ### Cross Platform Unit Tests