Nested geometric product error

```Julia
julia> @basis 1
(⟨+⟩, v, v₁)

julia> @basis 2
(⟨++⟩, v, v₁, v₂, v₁₂)

julia> A = Chain{V,1}(Chain(v1), Chain(v1))
(1v₁ + 0v₂)v₁ + (1v₁ + 0v₂)v₂

julia> x = Chain(v2)
0v₁ + 1v₂

julia> A ⋅ x
1v₁ + 0v₂

julia> A * x
ERROR: MethodError: no method matching Chain{⟨++⟩,1,Int64,2}(::Float64)
Closest candidates are:
Chain{⟨++⟩,1,Int64,2}(::T) where T<:Number at boot.jl:715
Chain{⟨++⟩,1,Int64,2}(::Simplex{V,0,B,T} where T where B) where {V, G, T, X} at /Users/eschnett/.julia/packages/Grassmann/3DicJ/src/multivectors.jl:44
Chain{⟨++⟩,1,Int64,2}(::Base.TwicePrecision) where T<:Number at twiceprecision.jl:243
...
Stacktrace:
[1] convert(::Type{Chain{⟨++⟩,1,Int64,2}}, ::Float64) at ./number.jl:7
[2] macro expansion at /Users/eschnett/.julia/packages/StaticArrays/1g9bq/src/util.jl:11 [inlined]
[3] convert_ntuple at /Users/eschnett/.julia/packages/StaticArrays/1g9bq/src/util.jl:8 [inlined]
[4] SArray{Tuple{4},Chain{⟨++⟩,1,Int64,2},1,4}(::Tuple{Chain{⟨++⟩,1,Int64,2},Float64,Float64,Chain{⟨++⟩,1,Int64,2}}) at /Users/eschnett/.julia/packages/StaticArrays/1g9bq/src/SArray.jl:28
[5] StaticArray at /Users/eschnett/.julia/packages/StaticArrays/1g9bq/src/convert.jl:4 [inlined]
[6] macro expansion at /Users/eschnett/.julia/packages/Grassmann/3DicJ/src/products.jl:0 [inlined]
[7] *(::Chain{⟨++⟩,1,Chain{⟨++⟩,1,Int64,2},2}, ::Chain{⟨++⟩,1,Int64,2}) at /Users/eschnett/.julia/packages/Grassmann/3DicJ/src/products.jl:856
[8] top-level scope at REPL[67]:1
[9] eval(::Module, ::Any) at ./boot.jl:331
[10] eval_user_input(::Any, ::REPL.REPLBackend) at /Users/eschnett/src/julia-1.4/usr/share/julia/stdlib/v1.4/REPL/src/REPL.jl:86
[11] run_backend(::REPL.REPLBackend) at /Users/eschnett/.julia/packages/Revise/0meWR/src/Revise.jl:1075
[12] top-level scope at none:0
```

Reproduce the nested Chain example in Julia, then inspect Grassmann's products.jl around line 856, where the stack trace reports the failing multiplication. Compare the working dot product with A * x and determine the expected geometric-product result; done means the multiplication no longer raises the shown MethodError and has regression coverage.

Written by the indexing model from the issue text.