what would be a good substrate?

The best substrates based on your notions, something that possesses porous grains, this allows each unit of the gravel to act independently having both aerobic and ananerobic regions on each grain. If the grain also have good Fe/Mn/Zn etc then even if the region around the grain is aerobic, the grain can still provide the reduce forms to plant roots.

Humus, soil etc is fine as well. As far CEC, I think it matters much less than so many tout. MPV turface vs Flourite is an example, turface is high in CEC and flourite low but I'd say Flourite grows better plants.

Folks want to talk about CEC like it's important but when you actually compare the character at issue, I've found it has little bearing on the claims.

Organic matter is good within reason. A balance should be struck when setting up a tank and the method you plan on using.

Slower growing lower light and/or non CO2 tanks work very well with a high% of soil, peat, mulm etc.

Using onyx sand or flourite over 1/2-1" soil or peat + mulm works very well over the long term for these types of tanks.

For higher light/CO2 tanks were replanting is more frequent, then less organic matter is used/needed generally.

1/4" of peat or a very thin layer + some mulm etc works well.

This starts off the tank's substrate quickly and after aboutn a month or two wears off. By then the tank's bacterial layers have formed. Peat keeps the substrate slightly reduced till the bacteria take over this role. The mulm adds what's missing from an old established substrate.

You can add sorts of concotions to the substrate but generally simple is good. I used plain sand + RFUG for a decade and fertilized the water column. Worked very good. Substrates come into play when you have less than optimal nutrients in the water column or cannot dose them in sufficent amounts for whatever reason.
They also become important in cycling and remineralization as they have enormous surface areas. This becomes more important in lower light/non CO2 methods while the high light/CO2 methods tend to rely more on inorganic dosing.

Regards,
Tom Barr