Fix FS#151  "$ns report" get's called in seemingly random places"

It turns out not to be random. Instead send_report is being called
when something fails in a timeline. The intent here (I assume) is to
inform the user of the error. Unfortunately this has several
undesirable side-effects such as creating an unnecessary zip file and
overwritten the existing contents of report.mail. I fixed this by
creating a new function send_error_report which is like send report
but doesn't have the undesirable side-effects.

statements.

by pid/id since the id is now an integer and not even shown in the
resulting drop-down list.

Also add plabnodehistmetrics, which ranks plab nodes using
libplabnodehist, writes the ranks and metadata to
/share/planetlab/reliable_nodes, and inserts a '+unavail' weight into
node_features for planetlab nodes.

utilities (tevc, linktest, libraries).

everybody's stuff up.

"select" to a "select distinct" on osidtoimageid since there will be
multiple osid entries for combo images.

event clients) need slightly different args.

no complete event to indicate an error condition.

it too, but its not like people use the page that much, so not gonna
bother.

at least under FreeBSD 6.2.  That is, a single kernel for regular, delay
and linkdelay usage.  I got close, still have a second kernel for delay
since it uses DEVICE_POLLING and I haven't had a chance to determine the
impact of the DEVICE_POLLING option when polling is disabled.

[Aside: polling + 10000HZ may be a bad idea.  In 6.2, there seems to be
 a lot more overhead associated with polling, they don't recommend you
 go over 1000HZ with polling turned on.]

Note that the script now checks the running kernel's HZ value, tweaks
/boot/loader.conf, and reboots if it is not correct.  However, this should
not be needed when combined with the latest, spiffy emuboot which can take
a HZ=NNNN command line option and change the kernel setting before loading
the kernel.

These changes should not affect the old 4.x kernels.

will give me a little breathing space (stage things in slower) by
starting just with boss.

there are things like images and osids owned by people in that group.
But exclude elabman/elabckup since they are created in the inner elab.

was turning out to be counterproductive on the ops db.

verbose option for debugging.

since it was written.