Allow active material volume fraction to be a function of space

[rtimms]

Currently we assume that the active material volume fraction eps_s is a scalar (but we allow the porosity eps to be a function of x, which in turn makes the inactive volume fraction eps_inactive a function of x by eps_inactive = 1 - eps - eps_s).

After #1115 the surface area per unit volume will be computed based on particle shape for li-ion (e.g. a = 3*eps_s/R for spheres), and currently you only get a(x) if you have R(x). Some of the leading-order models are only correct for constant a (e.g. leading order electrolyte conductivity), which is fine when eps_s is constant as they also only work for constant R. That is, it doesn't make sense to use SPM if R=R(x). These will need to be updated if we allow eps_s(x) which lets you have a(x) but with constant R.

You can also formulate a MPM (similar to SPM but with R(x), see e.g. https://www.sciencedirect.com/science/article/pii/S0013468620302541), but the formulation is significantly different for it to require new submodels, so would be a separate issue.

[valentinsulzer]

This has been fixed, presumably by #1262 but maybe earlier