eadp restraint wrongly converted from res on isotropic atom

[pascal22p]

EADP on isotropic atom is converted to U(IJ) restraint which is not supported. Should be using VIBRATION instead

[pascal22p]

depends on #26

[richardicooper]

See #26. EADP should be a constraint, RIDE will work to constrain iso or aniso in a compact instruction.
If restraint is required we could use a general restraint:
restrain 0.0, 0.01 = C(27,U[iso]) - C(13,u[iso])
For >2 atoms this becomes complicated because we wish to restrain each to the average, so lots of restraints explicitly needed (they are implicit in a U(IJ) with >2 atoms).