Defeat buffering in asan_symbolize.py

When piping Chrome stdout/stderr through asan_symbolize.py, we want it
to output every line immediately. Python line iterators buffer multiple
lines, so switch to using readline() directly instead.

Change-Id: I90d712c14094929e411b395d0a8d3ae00c0e9d05
Reviewed-on: https://chromium-review.googlesource.com/1114308Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: James Darpinian <jdarpinian@chromium.org>
Cr-Commit-Position: refs/heads/master@{#570286}

tools/valgrind/asan/third_party/asan_symbolize.py

@@ -436,7 +436,13 @@ class SymbolizationLoop(object):
 
   def process_logfile(self):
     self.frame_no = 0
-    for line in logfile:
+    # We use readline directly instead of an iterator ("for line in logfile")
+    # because the latter introduces extra buffering that is undesirable for
+    # interactive use.
+    while 1:
+      line = logfile.readline()
+      if not line:
+        break
       processed = self.process_line(line)
       print('\n'.join(processed))