Proposal on config for includeAttributeStatement

[Rod Widdowson]

> it doesn't cover the scenario in which you
> need to resolve attributes in order to produce a NameID, which is
> generally the case unless transients or database-backed persistents are
> used.

I'm not sure I understand why these are special - but I assume this springs
from the separation of NameID from attribute resolution work you did.

For the rest this makes sense to me.