The same-position question is key to resolving whether
these are random litho, or on-mask defects.
I can't see why those "stringers" would be where they
are and so consistently, in fabrication. Normally an
etch failure would leave residual metal at the bottom,
not halfway up. Is there a multi-step interlevel dielectric
involved, or is flat-field ILD thickness half of the metal
height? What does anything like SIMS say about
composition of this "feature"?