term matrix should report counts for genes in aspect that do not slim up

[cmungall]

No description provided.

[ValWood]

this will be a little complicated.

Normally we would want

1. Counts for genes that do not have any GO annotation to the specific aspect (root or none root)
2. Counts for genes which are annotated, but only to the root node for the aspect
3. Counts for genes which are annotated to the aspect, but are not in the submitted term list.

However, you can mix terms from different aspects in the Matrix tool (this is increasingly useful for annotation checking).
So these options would need to be restricted to cases where all the terms are from a single aspect