add a better error.

[thorstenhater]

Since we have support for ACC loading the errors thrown by paint are no longer
sufficiently informative. Even before that it was already hard to correlate source
and locus of painting errors.

Example

$> python3 main.C51A.py
Traceback (most recent call last):
  File "main.C51A.py", line 33, in <module>
    cell = A.cable_cell(morph, labels, decor)
RuntimeError: cable_cell: cable (cable 6 0 1) overpaints

This adds more details to the error message

$> python3 main.C51A.py
Traceback (most recent call last):
  File "main.C51A.py", line 33, in <module>
    cell = A.cable_cell(morph, labels, decor)
RuntimeError: cable_cell: Setting property 'ion-reversal-potential' on region '(region "soma_group")' overpaints at '(cable 6 0 1)'

[noraabiakar]

Looks good thanks! One small suggestion on what we can print.

[noraabiakar]

We have been printing everything in the decor part of the code as s-expressions, so maybe it's worth sticking to that style? Also we could print the value while we're at it.
Something like (init-membrane-potential -65).

[thorstenhater]

I deliberated on the same question, here's what I swayed me against

1. telling what might be enough information
2. ...and not overloading the reader, the line is quite long already.
3. also, where to draw the line? (density ('name' (...)))? Or print all the parameters?

[noraabiakar]

Fair enough. Should we keep it out of the public header as it's not printing the "full" information about the class?

[thorstenhater]

Ok, good point to think about. It just prints the outer DSL constructor in s-exp-speak.
At some point we had opened an issue on a 'general object printing facility', maybe it's
a good time to take that a bit seriously?

For now, I'll change << to print_paintable and hide it.

[brenthuisman]

Excellent.