[Geodata] [Tiger] non-road address ranges

[John P. Linderman]

Of the 837645 edges in New Jersey, I found 21 cases where an address
range was associated with an edge for which ROADFLG was not Y.

In every case
1) The FULLNAME of the edge was empty
2) The MTFCC of the edge was P0001, which, according to
    http://www.census.gov/geo/www/tiger/cfcc_to_mtfcc.txt
   corresponds to some sort of nonvisible boundary
3) The MTFCC associated with the address range was D1000,
   which doesn't even appear in the table (making it
   suitable for a nonvisible code, I suppose :-)

Case 1) got me wondering if there were any ROADFLG=Y edges
with address ranges but an empty FULLNAME.  Not in New Jersey,
anyway.  The edits mentioned at the bottom of page 3-74
(page 92 of 131) of TGRSHP07.pdf probably would have caught
this condition.

The "right thing" to do with any address range associated with
an empty FULLNAME is probably to ignore it.  I can supply a list
of ARIDs if we decide to go ahead with our "collection of
corrections".  -- jpl