Documentation nits

[mlubin]

Hi, a nit for the comparison with JuMP:

Modeling Interface: ExaModels.jl requires users to specify the model equations always in the form of Generators. This restrictive structure allows ExaModels.jl to preserve the SIMD-compatible structure in the model equations.

• There's nothing essentially different between generators and macros that JuMP uses; generators are syntactic sugar for a special case of abstract interpretation that macros do more generally.
• It's possible to write non-SIMD-compatible structures using generators, e.g.:

julia> c = ExaCore();

julia> x = variable(c, 10);

julia> A = [[1,2,3], [4,5]];

julia> constraint(c, sum(x[i] for i in A[j]) for j in 1:2)
ERROR: ArgumentError: invalid index: ExaModels.ParSource() of type ExaModels.ParSource

For this point I would focus more on SIMD restriction per se rather than generators.