gEDA-dev: pcb: paste time is non-linear

[DJ Delorie]

DJ Delorie <dj at delorie.com> writes:
> Ok, so I'm working on the challenge board, and wrote a perl script to
> panelize a hundred of them on a board.  Interestingly, each paste took
> longer than the previous one: starting at 29 pastes per second, up to
> about one second per paste in the end.

While fixing a different bug, I discovered the cause of this slowdown.
The "unique refdes" flag was on, so each paste required a renumber.

Sigh.