Make BackupTransport more resilient

Currently, if the transport implementation throws an exception, the
framework will wait for response for the full timeout duration (10
minutes) before aborting the operation (see attached bug).

1. Add exception handlers to the transport wrapper that runs on the
   remote side (i.e. transport implementation side) - while exceptions
   are no longer propagated to system_server with introduction of async
   transport, we'd still like to know when one occurs to stop waiting
   for response (instead of relying on timeout).
2. Reduce the 10-minute timeout to 5 minutes - our data shows this is a
   reasonable value.

Bug: 229670840
Test: 1. Manual (test B&R of any app)
      2. atest CtsBackupHostTestCases
Change-Id: Id97fa470fae65efc0ad587615a6e9edfd1e5f298