httpd threads max out

Some recent change brought in this bug. We observed that after a fresh Apache start, the httpd process count will keep increasing, and max out after a few hours. At this point any new api call such as /api/3/action/package_search will cause 500 error with apache error:
`Error - : can't start new thread`

Here is the detailed error info.

```
...
Module ckanext.googleanalytics.controller:65 in __init__
<< t = AnalyticsPostThread(self.analytics_queue)
t.setDaemon(True)
t.start()

def _post_analytics(
>> t.start()
Module threading:474 in start
<< _active_limbo_lock.release()
try:
_start_new_thread(self.__bootstrap, ())
except Exception:
with _active_limbo_lock:
>> _start_new_thread(self.__bootstrap, ())
error: can't start new thread
```

Start at ckanext.googleanalytics.controller line 65, where AnalyticsPostThread is created and started, and trace its lifecycle around _post_analytics. Reproduce the issue after a fresh Apache start while monitoring httpd process or thread counts, then exercise /api/3/action/package_search. Done means the thread count no longer grows without bound and the endpoint does not fail with “can't start new thread”.

Written by the indexing model from the issue text.