We need to be pragmatic about footprint growth in Gecko. A (n)k "rule" is silly. We need to understand that a requirement of embedding mozilla is that size matters - in everything that we write. So, if you’re a developer on the gecko part of mozilla - our client is embedders. There isn't anyone else -- and yes, eventually the Mozilla distributed client will be just another embedding customer.
If you add a feature, we should follow good software practices and aim to make the feature modular so that it can be optional. If you have a bug fix that does required a large footprint increase to fix, it may be a good idea to see if there is a better way to go about fixing the problem. It is finally up to the module owner to make a judgement call.
In the end, no one really cares about how great mozilla is, if it is so bloatly and monolitic that embedders can’t use it. Another better-built browser will takes its place. (Isn’t this the reason given by Apple when they chose kHTML over gecko?)