Copy input shape for batchedMatMulOp instead of append

[MarkKremer]

If you look a bit up in the diff of op_math.go you'll see outerX := x[:len(x)-2]. Taking a range from a slice and then appending to it will override the original slice instead of copying it into a new slice. Not very obvious but that's how it works. This change does a copy so that the shape of the input node doesn't get altered.

I looked into slice.Clone() but returning a slice to the pool that I appended to didn't feel right. Manually calling BorrowInts() creates a lot of clutter. It would be nice if there was a function to clone+append in the tensor package to do this. But for now I chose this approach. However, it isn't clear if outerX can contain more than 1 element. If it doesn't, then I could simplify it some more.

This is a result of #404 but don't close it yet. There are still some other issues I'm having that I'm still figuring out.

[coveralls]

Coverage increased (+0.2%) to 64.523% when pulling dd90e4c on MarkKremer:fix-batched-mat-mul-op into 3eb51b4 on gorgonia:master.

[MarkKremer]

#391 fixes this too. I propose I remove my fix from this branch but keep the test. Then you'll have some extra tests for your own fix. :)

[chewxy]

agreed.. let me do some finnacking around the gomod things. (#391 relies on a newer version of tensor)

[chewxy]

Will merge this into master once CI says go