Implementation of Einstein summation convention for multidimensional matlab arrays, where repeated indices sum over.

Supports multiple:
- Outer products
- Inner products
- Singleton dimensions
- Pages

Example:
A = rand(5, 1,4,8); B = rand(4,10,5 ); R = tensorproduct('jzgi',A,'gxki',B,'kjg');
Outer: 'i','j', Inner: 'k', Page: 'g', Singleton: 'x','z', size(R) = [10,1,5,8].

Inputs:
- ind_R : string with indices of the output
- A : (non-empty) (multidimensional) matlab array
- ind_A : string with indices of array A
- B : (non-empty) (multidimensional) matlab array
- ind_B : string with indices of array B

Version compatibility: This implementation makes use of Matlab built-in function pagemtimes,
introduced in Matlab version R2020b. To make use of this implementation in previous Matlab releases, comment/uncomment few lines in the body of tensorproduct_core, as specified in the routine.

The performance of tensorproduct can be assessed by calling the function tensorproduct_benchmark.