Thoughts on internal convolution:
	It could ALWAYS be defined as truncation times transient convolution.
	That would always avoid clutter.

Internal boundary conditions in 1-D seem to work.
	
Transient boundary conditions in 1-D seem to work.

Transient boundary conditions in 2-D seem to work.

Internal boundary conditions in 2-D seem to work.

Could introduce into regressions.
	1-D invint(), need to limit scales.
	2-D PEF estimator.
