Allow unwrapping of wrappers.

[thorstenhater]

Fixes #2254

[Helveg]

Ok so if I scanned the changes correctly, one could now arbor.membrane_capacitance(...).value and that would be a quantity object, so x.value.value would be the magnitude?

[thorstenhater]

Exactly. It's a bit awkward, but there's also the scale, which is a scalar (not a quantity) potentially varying along the dendrite. Thus, we have the split into (value=(magnitude, unit), scale=iexpr).

[thorstenhater]

@Helveg you could give the whole affair a review if it does what you need and we'll merge it.

[Helveg]

I don't have arbor in my software environment anymore, and no bandwidth really :(

[boeschf]

x.value.value looks indeed a bit awkward. Wouldn't it be more consistent with the canonical use of units if the magnitude/value can be queried by x.value and the quantity by x.quantity (i.e. x.value == x.quantity.value)? Not sure if quantity can be used as a property here, though.

[thorstenhater]

Yes, that seem like a good solution. As we use a base value and a potentially varying scale, how about

base :: double
base_as :: unit -> double
scale :: iexpr
unit :: unit

that should cover all the bases (!!).

[boeschf]

yes, better still. should it be base or basis?

[thorstenhater]

I wonder about that, too, but I tend towards base (as in 'base ten').