No subject

your region borders are way out in space. 

With following values (Wagner II) I don't see the latitudal wraps 
(assuming here a resolution of 10000m):

north:      9280000
south:      -9280000
east:       18560000
west:       -18560000
cols:       3712
rows:       1856
e-w resol:  10000
n-s resol:  10000

There are really no good guidelines published on how you should set the
NSEW borders for locations in different projections in GRASS. Here (as I
usually do) I made the distances west-east and south-north even multiples
of the ew_ and ns_ resolution, respectively, but also even multiples of
the number of columns and cells, respectively. (And multiples of 8 as
well, but that is just my preference and may not be important).

I arrived at these values through:
proj +ellps=WGS84 +proj=wag2
0 90
0.00    9265609.63
180 0
18531288.84     0.00

and then took the even multiples slightly larger than the maximum 
northing/easting values given by proj.

I have no problems with bordwalk() cropping the results.

Does this, with the changes you suggested to PROJ, solve it?

Morten Hulden