Changes in revision 1.218 caused reclaim_child_processes() to be too

quick in checking for child exit and sending additional SIGTERM and
SIGKILL interrupts.  Not a problem for systems with fast process context
switching, but Solaris complains bitterly.  The fix is to insert a few
quiet checks of the child status so that the exponential backoff on the
wait time becomes significant before sending additional interrupts and
complaining to the error log.

Reviewed by:	Dean Gaudet


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@79339 13f79535-47bb-0310-9956-ffa450edef68