In theory, these two things should yield identical results:

vecstc = apply_inverse(..., pick_ori='vector')
vecstc.plot(..., brain_alpha=1., overlay_alpha=1., vector_alpha=0.)

versus

stc = apply_inverse(..., pick_ori=None)
stc.plot(..., surface='white')

In practice, plot_vector_source_estimates calls brain.scale_data_colormap under the hood, while plot_source_estimates does not, so the automatic choosing of clim is quite different in the two cases. If you pass in a clim dict with your own values, then the two approaches indeed look identical, so probably (?) all that is needed is to remove the call to brain.scale_data_colormap.