2-stage duplication marking

[Henrik Nilsen Omma]

Hi,

I'm trying out a new way of tracking duplicates for tricky packages.

Marking duplicates is a good way of reducing the sheer volume of bugs 
but it can also be hazardous because marking dupes based on their 
symptoms might result in dropping bug reports that relate to different 
causes.

I've been scratching my head over some of these, as I'm sure you have 
all done as well. I'm thinking we should try to combine some of that 
pondering in a more structured way.

So I propose setting up wiki pages for potential duplicate farms on 
large packages. I've started with ubiquity: 
https://wiki.ubuntu.com/DebuggingUbiquity/DupCandidates

In that case you can see that I've actually set up a dupe farm for a 
rather weak case, where we still don't understand what is wrong. I've 
cleared it with the developer and I'm linking back to the wiki page from 
the bug to explain the reasoning.

I could imaging categories starting out as weak candidates for dupes and 
having the case strengthen as more people compare the cases, we get logs 
and the maintainers have a look. Once we are fairly sure we can mark the 
dupes in Malone.

Oh, and these pages can of course be used as sources for the info files 
in bughelper eventually as well.

Henrik