🚀 go-pugleaf

[more a question than a bug report, because it's not clear whether imageto
is expected to handle the following...]

Is imageto supposed to be able to separate glyphs that overlap in one
dimension?  I know it can't handle 2D overlap (where images actually
touch), but an image like

                 XXXX  YYYYYYY
                 XXXX      YYY
                 XXXXXXXXX YYY
                 XXXXXXXXX YYY
                 XXXX      YYY
                 XXXX  YYYYYYY

is yielding glyphs

                 XXXX
                 XXXX
                 XXXXXXXXX
                 XXXXXXXXX
                 XXXX
                 XXXX

and

                 YYYYYYY
                     YYY
                 XXX YYY
                 XXX YYY
                     YYY
                 YYYYYYY

namely the "left" glyph is correct, but the "right" glyph includes those
portions of the former which are contain in the latter's bounding box.  Am
I doing something wrong, or is this a known deficiency in imageto?  If the
latter, I would think it would be a real pain to deal with italic/slanted
faces, which probably overlap BB's a lot...

I'm also getting some (related?) oddities for the two double quote glyphs.
It's kind of hard to describe in ASCII (I can try and isolate the bits in
question if you want to see them), but IMG data something like

               xx     xx
              xx     yx
             xxx    zyx
             xxxxx  zyxxx
             xxxxxx zyxxxx
              xxxx   yxxx

is being converted by imageto into a GF char looking like

               xx     xx
              xx      x     y
             xxx      x    zy
             xxxxx    xxx  zy
             xxxxxx   xxxx zy
              xxxx    xxx   y

i.e. part of the left side of the rightmost segment is getting moved to the
right side.  The rightdblquote is similarly afflicted, but no others.  I've
noticed that when running -verbose, imageto outputs a plus-sign ("+") in
front of just these two id's; no others.  What's the meaning of the plus
sign?  Are these two segmentation problems related?

thanks for any help you can offer,
Pete