Bug in pofile-daily-stats

Jeroen was jealous that I decided to go for the "proper" fix for bug 409330 with bug 411327, so he spent some extra time looking at my code. And found a nasty little bug and mentioned it in the merge proposal, but unfortunately, only after the code has already landed.

So, if anyone thinks jealousy is bad, think again!

Here's Jeroen's comment from https://code.edge.launchpad.net/~danilo/launchpad/bug-411327/+merge/9913:

> I had one remark:
>
> + OtherPOTJoin = Join(
> + OtherPOT, And(OtherPOT.name == MatchingPOT.name,
> + OtherPOT.id >= MatchingPOT.id))
>
> This is what we normally do to reduce duplicates in joins where two
> different rows of the same table can be joined symmetrically. But is it
> really justified in this case, where the two POTemplate rows are not in
> symmetrical roles?
>
> With this clause, AFAICS the script updates only the POFiles from newer
> POTemplates than the one that was recently updated. Is that
> intentional? As an example, I can break
> test_POFileSet_getPOFilesTouchedSince_shared_in_product by changing the
> order in which its two templates are created.