Make field names from sources more descriptive

The VTK-m sources (like Oscillator, Tangle, and Wavelet) were all creating fields with very generic names like pointvar or scalars. These are very unhelpful names as it is impossible for downstream processes to identify the meaning of these fields. Imagine having these data saved to a file and then a different person trying to identify what they mean. Or imagine dealing with more than one such source at a time and trying to manage fields with similar or overlapping names.

This changed also moved the cellvar output out of the Tangle source. This field had nothing to do with the tangle computation. Rather, it was a simple counting array echoing the cell indices, which was confusing. Instead, this functionality has been moved to a new GenerateIds filter. This filter adds a pair of fields to a DataSet which mirror the indices of the points and cells, respectively. These fields are useful for tracking the provenance of the elements of a DataSet as it gets manipulated by the filters. It is also convenient for adding indices to operations designed for fields and for testing purposes.