Marking move from one component type to another one as impossible for schema evolution

[hegner]

BEGINRELEASENOTES

• Detect more impossible schema evolutions
• Improving printout of schema evolution checker ENDRELEASENOTES

This addresses #563

[hegner]

This turned out to be a slightly simpler change than what I would have imagined :).

Probably speaks for the design for once :-D

[hegner]

Can we add a simple test that at least runs the script on the example datamodel(s) that we have?

Surprised we didn't have. This PR or a new one?

[tmadlener]

Yeah, I also thought we had one, but at least I couldn't find one. Or if we have one at least it didn't trigger this issue. I would prefer a new PR.

[hegner]

We only had for the C++ part it seems. OK. new ticket, new PR

[tmadlener]

PR to fix the bug in the schema evolution generation: #627

[hegner]

other PR merged now

[tmadlener]

Can you rebase this quickly to pick up those changes?

[hegner]

obviously that fails now ;-) need to change the test definition

[hegner]

for post v1 we could do that a bit separated without a file, but in pure python tests. I will merge now.