Goldilocks and the AC timeout value

[Dan Scott]

See the thread at http://libmail.georgialibraries.org/pipermail/open-ils-dev/2011-June/007274.html for full details, but long story short: an AC timeout value of '1 second' was resulting in many added content requests timing out, but the subsequent change I made to '30 seconds' significantly increased the risk of blocking all Apache processes in the event of a malfunctioning added content provider or a malicious attack.

Therefore, in the Evergreen working repository, the user/dbs/saner-ac-timeout branch offers a change of the timeout value from 30 to 3, and additional comments describing the risks and rewards of changing the setting.

This should be able to be applied from master through rel_2_0.

[Mike Rylander (mrylander)]

Signed off and pushed to master through rel_2_0.