ftrace: Turn TraceCollector into a context manager

This allows using an TraceCollector instance as a context manager, so
tracing will stop even if the devlib application is interrupted, or if
an exception is raised.

devlib/trace/__init__.py

@@ -31,5 +31,13 @@ class TraceCollector(object):
     def stop(self):
         pass
 
+    def __enter__(self):
+        self.reset()
+        self.start()
+        return self
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self.stop()
+
     def get_trace(self, outfile):
         pass

doc/overview.rst

@@ -323,17 +323,14 @@ You can collected traces (currently, just ftrace) using
    # the buffer size to be used.
    trace = FtraceCollector(t, events=['power*'], buffer_size=40000)
 
-   # clear ftrace buffer
-   trace.reset()
-
-   # start trace collection
-   trace.start()
-
-   # Perform the operations you want to trace here...
-   import time; time.sleep(5)
-
-   # stop trace collection
-   trace.stop()
+   # As a context manager, clear ftrace buffer using trace.reset(),
+   # start trace collection using trace.start(), then stop it Using
+   # trace.stop(). Using a context manager brings the guarantee that
+   # tracing will stop even if an exception occurs, including 
+   # KeyboardInterrupt (ctr-C) and SystemExit (sys.exit)
+   with trace:
+      # Perform the operations you want to trace here...
+      import time; time.sleep(5)
 
    # extract the trace file from the target into a local file
    trace.get_trace('/tmp/trace.bin')