netsniff-ng may be HUP'ed incorrectly if the PCAP processing directory is backlogged

[mmguero]

This is very similar to issue #34.

The pcap-capture/scripts/netsniff-roll.sh script which is executed in the context of the pcap-capture container is responsible for making sure that netsniff-ng is HUP'ed if the PCAP file it is writing to exceeds the value for the PCAP_ROTATE_MINUTES environment variable (the PCAP_ROTATE_MEGABYTES value is actually handled by netsniff-ng itself with the -F argument).

However, if the pcap/upload directory is very full, the same files will be viewed multiple times. This will cause the thing to hup netsniff-ng every few seconds.

Possible fixes:

• capture to a different directory than the uploads (does slightly complicate things)
• only HUP netsniff-ng for the most recent PCAP file per-process, rather than all of them (probably the easiest solution?)
• only HUP netsniff-ng for PCAP files that are newer than the last time we hupped that process (also probably pretty easy)