Compatibility with datasets without N-dimensional form

[ssomnath]

Test all applicable functions with:

1. Incomplete measurements (position axis)
2. Sparse sampling (position axis)
3. BEPS relaxation - dummy spectroscopic dimension

Thinking about guiding the self-explanatory handling of such datasets. Perhaps add attributes to the HDF5 file letting the code know how to handle the special case datasets.

One suggestion - add the following attributes to the ancillary datasets:

1. dependent_dimensions : ['Field'] - this will handle dummy spectroscopic dimensions
2. incomplete_dimensions : ['X', 'Y'] - this will handle both sparse sampling and incomplete measurements

Upon noticing either of these attributes, functions such as reshape_to_n_dims() would ignore all dimensions in dependent_dimensions when reshaping and collapse all dimensions in incomplete_dimensions to a single dimension in a numpy / Dask array

[ssomnath]

E.g. - in FORC BEPS SPM - DC_Offset indices never get reset since values are NOT the same per FORC cycle. Pycroscopy's BELoopFitter works around this issue by working entirely on the 2D dataset rather than reshaping to N-dimensions