sympy backend as of JuliaCon 2020

This is SymbolicTensors.jl with a few fixes and new general relativity functions from how it was presented at JuliaCon 2020.
Much of the tensor calculus is still done in sympy.tensor.tensor via SymPy.jl, though some (e.g. taking derivatives) is done in Julia. The master branch and newer releases will eliminate this sympy dependency in exchange for SymbolicUtils.jl, as well as introducing breaking changes to the user interface.