Support for templatebrains that share same physical space

[jefferis]

See e.g. natverse/nat.flybrains#40 for the kind of situation where this might be useful. We could add a field in the templatebrain constructor for a parent template e.g.

JRC2018F and JRC2018FDS with the latter pointing to JRC2018F in a field called something like parent or base.

It could also help e.g. to have template objects for calibrated / uncalibrated versions of the same space (although ideally I suppose one would avoid the clutter of additional templates). I can basically think of two options

1. Adding a child template (as above)
2. wrapping one of the endpoint templates (i.e. reference or sample in xform_brain) in a function that indicates that it is uncalibrated e.g.

xform_brain(x, sample=uncalibrated(JRCFIB2018F), reference=FCWB)

One would need to augment the bridging graph with 0 length edges between the uncalibrated templates and the calibrated versions.