HDF5 dims labels are incorrect for vector quantities

[jsoishi]

Vector quantities output to HDF5 files are missing the dimension corresponding to vector/tensor component(s). Instead they currently skip that dimension, moving to the next one and leaving the trailing dimensions blank. Here's an example from a 2D velocity field.

u_dset = df['tasks/u']
for i in range(len(u_dset.shape)):
    print("dimension {} label: {}".format(i,u_dset.dims[i].label))

[kburns]

With #196, the tensor axes are now skipped (no scales) so that the spatial scales are at least attached to the correct dimensions.