I agree about the problem, and think you explain it well.  In fact, many academic studies are using somewhat better measures for the reasons you describe; my preferred measure is the per capita value of oil/mineral production.  Still, I have some sympathy with institutions like the World Bank that use other measures to identify countries that are 'resource rich' (but really meaning, as you imply, 'resource dependent'), since they are often those for whom extractive sector reforms would be most salient.

Agreed - Keith Myers at Chatham House wrote a useful article on oil resources per capita a few years back - have a look at Myers, K., Chatham House Briefing Paper – Petroleum, Poverty and Security,  London, 2005 – available at http://www.chathamhouse.org/sites/default/files/public/Research/Africa/bppetroleum.pdf

My rough maths has Nigeria producing 5.5 barrels of oil per person per year, whereas New Zealand produces not much less at 4.4 barrels.  The former is the 'poster child' for the resource curse; the later never mentioned on the resources front. 

But one vote in favour of the IMF definition - the message that it sends by focusing on GDP / revenues / exports is that natural resources produce disproportionate risks in the absence of other economic sectors that might balance them out, and that correspondingly they are probably quite useful as part of the mix of a diverse economy.