VB, 9/3/2017:

The "intermediate" defaults are not yet finished but those that are
included here have seen some use. It is not clear that they are
"optimal" in some sense, and further tweaks to the intermediate
settings could occur as a result of further testing.

At least for elements 1-18, the present recipe to construct
"intermediate" settings should be straightforward to apply also for
some missing elements.

What I need help with is further testing. To create further
"intermediate" settings for different elements, what should be looked
at (compare to results for "tight" settings) is:

- Change of the dimer binding energy curve, LDA and HSE06 

- Change of total energy of a nontrivial monoelemental or hydrogenated
  molecule or cluster (several atoms), LDA and HSE06

- Change of the total energy of the elemental solid, LDA and HSE06

- Change of the lattice parameter and bulk modulus of the elemental
  solid, LDA and HSE06

- Change of the total energy, lattice parameter and bulk modulus of a
  simple bulk oxide, LDA and HSE06
  Steps:
  * compute for "tight" settings
  * compute change for "element, tight" + "O, intermediate"
  * compute change for "element, intermediate + "O, intermediate"

