solve_multivar might return not all solutions

[sumiya11]

Our solve_multivar makes assumptions about the transcendence basis which do not hold in this case:

julia> @variables x y z
julia> symbolic_solve([(x - y) * (x - z) * (y - z)], [x,y,z])
2-element Vector{Dict{Num, Any}}:
 Dict(z => z, y => y, x => y)
 Dict(z => z, y => y, x => z)
# missed y = z