Initial check-in of a perl module representing persistent leases.  Mostly
done.  Needs testing and a schema update to go along with it.

like SA.

the close (child has exited) and we hang out forever. Not sure why
this happens but probably related to all of the file descriptor
messing about. Needs more investigation.

We need to know if a blockstore is persistent so we can avoid blindly
doing a mkfs on those!

If camcontrol fails, it will still spew out info that will tells us
the mapping we need. But it will not be on the first line of output
so parse em all!

chain in the DB, which causes the samecert check to fail oon
pre-existing slices.

the nodecheck to run in MFS even if not in the reloading state.
if proj/<pid>/nodecheck/`uname -s`/bin-`uname -m`"
exits then add it to path and use the helper programs located there.

creators to active users list.

Yeah, yeah, I know I shouldn't update these files, but I am going
to gamble that no one has applied this change yet, rather than creating
yet another new update.

We put the lease in this state while allocating the storage associated
with a dataset lease.

Upstream pulled in a patch to include this, so we no longer need
a local patch to gitlab - yay! However, they put it in a slightly
different place in the webhook data than I did.

Please let this be the last commit of any significance for a while
for the force10 module.  I discovered that tricky bitmath was required
to get the port removal right when mixing tagged and untagged ports.

It's useful to know what state an interface is in currently versus what
is mandated in the 'interfaces' table.  E.g., when deciding what to do
in setPortVlan() based on whether or not a port is in 'trunk' mode.

* Clean up and normalize messages (warnings, errors, debug, etc.)
* Avoid DB lookups in convertPortFormat()
* Convert some direct snmp calls to library calls
* Zap trunk accounting code (via switch) - use Port state instead.
* Bit of code de-dup.

Added port trunking methods to the snmpit_force10 module, but
stopped short of adding port channel (LAG) functionality.  The latter
requires support from either a future REST API, or a somewhat hokey
HTTP-RPC mechanism that invokes switch-side scripts.

Also did some cleanup and minor refactoring.