api services need better error handling on startup

[Pavlo Shchelokovskyy]

Recently when deploying Heat on CentOS 6 (Python 2.6) we got following errors for all api services due to missing config files:

2014-09-12 09:59:27.461 2383 CRITICAL heat [-] UnicodeError: Message objects do not support str() because they may contain non-ascii characters. Please use unicode() or translate() instead.
2014-09-12 09:59:27.461 2383 TRACE heat Traceback (most recent call last):
2014-09-12 09:59:27.461 2383 TRACE heat File "/usr/bin/heat-api", line 64, in <module>
2014-09-12 09:59:27.461 2383 TRACE heat sys.exit("ERROR: %s" % e)
2014-09-12 09:59:27.461 2383 TRACE heat File "/usr/lib/python2.6/site-packages/heat/openstack/common/gettextutils.py", line 333, in __str__
2014-09-12 09:59:27.461 2383 TRACE heat raise UnicodeError(msg)
2014-09-12 09:59:27.461 2383 TRACE heat UnicodeError: Message objects do not support str() because they may contain non-ascii characters. Please use unicode() or translate() instead.

The actual error in fact was

Unable to load heat-api from configuration file /etc/heat/api-paste.ini.
Got: ImportError('No module named keystonemiddleware.auth_token',)

but was masked by the traceback above.

Looks like we should "sixify" error handling in heat-api-* scripts to better present errors to deployers.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/121473

[OpenStack Infra (hudson-openstack)]

Change abandoned by Pavlo Shchelokovskyy (<email address hidden>) on branch: master
Review: https://review.openstack.org/121473
Reason: yep, is a dupe