Wikipedia defines ecological footprint as the amount of biologically productive land and sea area necessary to supply the resources a human population consumes. Which overshoot can I talk about if I consume even a bit of non-renewable resources? How much biologically productive land and sea area are necessary to supply 1 barrel of oil or one tonne of aluminium?
I think you have the right idea. Your question addresses exactly what the ecological footprint does not measure, and the question you link to is trying to explain that.
Fundamentally, ecological footprint is about energy. Sunlight arrives and is used by plants (or a tiny amount is used directly by people via PV or other solar power systems), which use it to convert CO2 out of the air into more complex carbon molecules. So you can take a complicated cycle like "plants plus time to coal, coal plus iron to steel becomes car" and work backwards to say that the car took X amount of coal, or X amount of energy, to make.
As the Ecological Footprint measures the area required to produce a material or absorb carbon dioxide emissions, materials such as mercury that are not created by biological processes nor absorbed by biological systems do not have a defined Ecological Footprint
To use aluminium as you do, what they're saying is that aluminium as a metal does not have an ecological footprint, since nothing in the earthly environment makes aluminium. We have a finite supply of aluminium-containing rocks that we can refine the metal out of. Aluminium is not really a good example, since it's very common and not very toxic.
There's no direct ecological process that takes aluminium out of the environment. Nothing eats it, in other words. What happens is that it eventually oxidises and breaks up, forming inert compounds, in other words sand/rocks. This is perhaps most obvious with iron/steel, where parts of the world have naturally occurring iron ore right on the surface. In a lot of Western Australia they call it "dirt", because the soil is 99% iron ore. It's called haematite as a rock, or rust if you see it on an iron object. That cycle is fairly obvious: take rust (haematite), heat it and remove the oxygen and you get iron. Expose iron to air and water and it breaks back down to rust. Most chemicals have more complex pathways and are often toxic through the whole pathway - dioxins and plutonium, for example, take a very long time to break down and are nasty at almost every step. And there's really no way to express that in terms of carbon dioxide emissions.
What the ecological footprint can measure is the cost of finding aluminium ore (rocks), and turning those into useful aluminium metal objects. Or recycling them, or immobilising them to make them less poisonous. In essence, the processes are only about energy, and energy is relatively easy to express in carbon footprint terms.
Good question. There are three parts to the answer: extracting and using non-renewable resources also uses energy, water, etc which are part of the ecological footprint; "consuming" some non-renewable resources can have other negative effects on the environment; and there are limited amounts of non-renewable resources.
The footprint FAQ you reference is addressing the first point. To take Aluminium as an example, producing Aluminium from the ore requires a large amount of energy: the ore is heated to 1000°C and electrolysed using thousands of Amperes of current, with large amounts of CO2 given off in the process (see Hall-Heroult process). So there is a significant carbon footprint to its production (231 MJ/kg embodied energy according to this, plus direct CO2 emissions). Aluminium ore also tends to be strip-mined, which takes up land which could otherwise be farmed, and so on. These impacts should be included in ecological footprint calculations.
The second point is rather complex. The extraction and use of some resources can pollute the environment (e.g. Cadmium released from spent batteries, mercury released from gold mining). In theory this could be included in an ecological footprint calculation: the amount of land/sea needed to degrade the pollution plus the effective reduction in global capacity from the negative effects of the pollution. But in practice this calculation is difficult and doesn't seem to be included in most methodologies for measuring ecological footprint (e.g. see here). Also some kinds of pollution (e.g. plastic waste) just aren't degraded. One way of attempting to measure the most significant negative impacts of activities is the idea of Planetary Boundaries - which overlaps a bit with ecological footprint but also includes biodiversity, pollution, etc. Unfortunately this is less well-developed than ecolocigal footprint, and isn't very quantitative - yet.
Finally, in using a non-renewable resource you may be reducing the total amount available. I say "may" because many non-renewable resources are widely recycled (e.g. Aluminium). This scarcity angle is actually rather well handled by economics: price movements do a much better job of managing the use of scarce raw materials than they do of managing environmental impacts. So I tend not to worry too much about this on the assumption that scarcity is reflected in price. If only I could make the same assumption about environmental and social impact...