fixing op2 subcase id

[SteveDoyle2]

Takes advantage of the fact that:

• statics can have nsubcases (NSUB)
• modes can have 1 subcase, but NSUB modes
• craig-bampton is like modes and has 1 subcase and NSUB modes
• buckling has 2 subcases with NSUB+1 (static+modes) therefore we explicitly call out the subcase_index for modes/craig-bampton and buckling as 1 or 2 depending

corrected eigenvector table flag (should be 7 and not 1, which would lead to displacements being output)

fixes: #31

Much better output

# test_op2 EB-ALL-ELEM-TEST-LANCZOS-MODE-2-DPB.OP2
eigenvectors[1]
  isubcase = 1
  type=RealEigenvectorArray ntimes=2 nnodes=13, table_name=OUGV1
  data: [t1, t2, t3, r1, r2, r3] shape=[2, 13, 6] dtype=float32
  node_gridtype.shape = (13, 2)
  sort1
  modes = [1 2]; dtype=int32
  eigns = [46891.77  61105.016]; dtype=float64
  mode_cycles = [216.545 247.194]; dtype=float64

# test_op2 EB-ALL-ELEM-TEST-GIV.OP2
eigenvectors[1]
  isubcase = 1
  type=RealEigenvectorArray ntimes=4 nnodes=13, table_name=OUGV1
  data: [t1, t2, t3, r1, r2, r3] shape=[4, 13, 6] dtype=float32
  node_gridtype.shape = (13, 2)
  sort1
  modes = [1 2 3 4]; dtype=int32
  eigns = [ 46891.766  61105.016 156801.047 216221.312]; dtype=float64
  mode_cycles = [216.545 247.194 395.981 464.996]; dtype=float64