Clifford2 inverse is still wrong.
It passes the tests but it has code that could fail when an inverse exists.
I suggest here that it may be possible to define a function f such that for every Clifford number x, (* x (f x)) is either zero or one.
For a zero divisor the cofactor is thus displayed.
I assume further that this can be recursively defined.
Perhaps a bool should also be returned.
This is reminiscent of the Penrose general matrix inversion.
Here is a test which seems to refute the idea.