all liquids is a fine measure of oil production, except it isn't

Don't be silly.

The non-oil components of "all liquids" compete directly with oil in a number of markets - NGL in gasoline blending components and ethylene feedstocks (plastics), for example - so looking at all liquids is absolutely the correct measure.

It might make sense to normalize the volumes for energy content, but then again it might not - it's not clear that different energy levels make a difference for feedstocks.

What absolute nonsense. Even if they do compete directly, which is only true to a degree, they don't all have the same energy density, so volumemetric comparisons are meaningless. Just assuming that a barrel of oil is equivalent to an barrel of ethanol is plain wrong.

Even if they do compete directly, which is only true to a degree, they don't all have the same energy density

Does that matter for ethylene feedstocks?

I have no idea, but I suspect neither do you, so there's no indication energy density is the key factor you suggest. Hence my qualification that normalizing by energy density may or may not be wholly appropriate.

At any rate, the point is simply that non-crude liquids in the oil supply directly substitute for crude in some very substantial uses. Accordingly, an increase in those other liquids displaces crude use in those areas and makes more available for other uses, effectively increasing the supply of crude. We can quibble about the relative worth of different liquids (I suspect NGLs and ethanol should be discounted by 20-30%), but that's a secondary issue to the fact that the EIA includes all liquids in "total oil supply" for a very good reason.