Second attempt, using the "float" technique instead of the "wipe"
technique (same url, more data):
http://www.delorie.com/pcb/brass/Net result: even with all the "worst case scenario" things (poor toner
coverage, partially spent FeCl, long etch times) the results are
better than yesterday (good coverage, fresh FeCl, short etch time).
So, float wins over wipe for brass etching.
Still haven't tried applying paste through them yet. Hopefully
tomorrow.