Make SyclTimer accumulative

```

In [9]: timer = dpctl.SyclTimer()

In [10]: with timer(q):
    ...:     y = dpt.linspace(1, 2, num=10**6, sycl_queue=q)
    ...:

In [11]: timer.dt
Out[11]: (0.0022024469999450957, 0.002116712)

In [12]: with timer(q):
    ...:     x = dpt.linspace(0, 1, num=10**6, sycl_queue=q)
    ...:

In [13]: timer.dt
Out[13]: (0.004531950999989931, 0.004239664000000001)
```

Author: oleksandr-pavlyk

dpctl/_sycl_timer.py

@@ -67,10 +67,8 @@ def __init__(self, host_timer=timeit.default_timer, time_scale=1):
         self.timer = host_timer
         self.time_scale = time_scale
         self.queue = None
-        self.host_start = None
-        self.host_finish = None
-        self.event_start = None
-        self.event_finish = None
+        self.host_times = []
+        self.bracketing_events = []
 
     def __call__(self, queue=None):
         if isinstance(queue, SyclQueue):
@@ -89,27 +87,30 @@ def __call__(self, queue=None):
         return self
 
     def __enter__(self):
-        self.event_start = self.queue.submit_barrier()
-        self.host_start = self.timer()
+        self._event_start = self.queue.submit_barrier()
+        self._host_start = self.timer()
         return self
 
     def __exit__(self, *args):
-        self.event_finish = self.queue.submit_barrier()
-        self.host_finish = self.timer()
+        self.host_times.append((self._host_start, self.timer()))
+        self.bracketing_events.append(
+            (self._event_start, self.queue.submit_barrier())
+        )
+        del self._event_start
+        del self._host_start
 
     @property
     def dt(self):
         """Returns a tuple of elapsed times where first
         element is the duration as measured by the host timer,
         while the second element is the duration as measured by
         the device timer and encoded in profiling events"""
-        self.event_start.wait()
-        self.event_finish.wait()
-        return (
-            (self.host_finish - self.host_start) * self.time_scale,
-            (
-                self.event_finish.profiling_info_start
-                - self.event_start.profiling_info_end
-            )
-            * (1e-9 * self.time_scale),
-        )
+        for es, ef in self.bracketing_events:
+            es.wait()
+            ef.wait()
+        host_dt = sum(tf - ts for ts, tf in self.host_times) * self.time_scale
+        dev_dt = sum(
+            ef.profiling_info_start - es.profiling_info_end
+            for es, ef in self.bracketing_events
+        ) * (1e-9 * self.time_scale)
+        return (host_dt, dev_dt)