1. I suggest to add some function filling the tensor (because sth must produce the tensor from input data)
2. Do we allow to use such function as memcpy or not?
   (it would be faster, but I already have the implementation without it if there would be a problem)
3. If we try to add tensors with different size, there should be an error or not?
   (I could implement to return a longer tensor, in subtraction -> always first but the loop could be in its size)
4. What about create fastGetTensorField where index is argument, not a Point?
   (it is sth for loops, where we need access to all index, and don't want to count it every time)
5. Only double fields or void*?

1. initVector/1, pushBackVector/2:
   a) we want to know the size at the beggining and add only values
   (don't know if all elements were initialized)
   b) or create 0 size vector and then add elements
   (don't know if we got the proper size)
   Prefer: b)
2. error report
   How to report problems? The printf warning is enough or create special "error returning value"?
3. Do we need any kind of deleting vector data fields?

Please do not change my code. Work is still in progress. I just need the answers to continue.

I am thinking about:

1. adding Point copyPoint(Point point) function (needed in tensor init function and maybe somewhere else)
2. adding Point initPoint(int x, int y, int z) function (because file with only copyPoint function will be sad)
3. rename x, y, z to something more adequate