consecutive means the tail is within the same subelement inames as
the vectorized loop (except for the vectorized iname)
blocked means each tail has its on set of subelement inames

odd behavior for unstructured tensor test...

Unix has a hard 256 character limit on filenames. I have no idea
how to work around this except for applying a hash function.

The permutation is already in. Using the numbers requires to have the
correct global state whenever the key is evaluated, which we cannot
guarantee.

The unstructured stuff changed the implementation of the parallel key
to depend on the quadrature tuple. This means that we need to make sure
that the state of the quadrature tuple is correctly all the time!