Log entry time 12:19:09 on December 18, 2010

Entry number 345002

This entry is a followup to: 344792

The script end_clean was slightly modified: the caget is killed at
the end, also there is a "sleep 12" to delay this.

The solution in halog 344792 is recommended.

Ultimately I will need to solve this (of missing of some end-of-run halog
entries)  It was always a problem, but less for earlier experiments
because we had discouraged having very long end-of-run scripts.  
DVCS exposed the problem because their script is very slow (tons of 
EPICS HV being recorded).  However, the design will need to be 
reconsidered during the X-mas break.