Support Neurolucida multiple soma stack

[wizmer]

As in:

  (Color DarkMagenta)
  (CellBody)
  (   -3.19    -9.26     0.08     0.17)  ;  1, 1
  (   -3.53    -9.60     0.08     2.35)  ;  1, 2
  (   -1.01   -10.61     0.08     2.35)  ;  1, 3
  (    0.50   -11.11     0.08     2.35)  ;  1, 4
 )  ;  End of contour

("CellBody"
  (Color Orange)
  (CellBody)
  (    5.88     0.84    54.62     2.35)  ;  2, 1
  (    6.05     2.53    54.62     2.35)  ;  2, 2
  (    6.39     4.38    54.62     2.35)  ;  2, 3
  (    5.55     5.05    54.62     2.35)  ;  2, 4
  (    2.86     5.72    54.62     2.35)  ;  2, 5
)  ;  End of contour

("CellBody"
  (Color RGB (255, 4, 255))
  (CellBody)
  (    1.85     0.67   -90.22     2.35)  ;  3, 1
  (    0.84     1.52   -90.22     2.35)  ;  3, 2
  (   -4.54     2.36   -90.22     2.35)  ;  3, 3
  (   -6.55     0.17   -90.29     2.35)  ;  3, 4
) ; End of contour

[mgeplf]

I'm not sure there is a definite standard for this. I think z is
constant comes from this:
https://github.com/neuronsimulator/nrn/blob/99a6a370ef143b3c681eda1deea68bc7e2b8f21d/share/lib/hoc/import3d/read_nlcda3.hoc#L276

However, in the only other multi-CellBody reconstructions I could
find, the z wasn't constant (unfortunately, I can't share the files
here).

The neuromorpho soma spec mentions multi-contours here:

Refering to this file, but it's not in ASC, but in the DAT format, so I don't know what the ASC version looks like.
http://neuromorpho.org/neuron_info.jsp?neuron_name=AKO60sdax2lay

Until we get a definitive example, I'm going to leave this hanging, along w/ the potential solution #306