Some comments on the invariant plane strain approach to martensite crystallography