[develop] Descriptor validation for miniscript

[tadeubas]

Descriptor validation for miniscript branches was not working properly

[odudex]

I think to better reflect real cases, test descriptors shouldn't have repeated keys. When they do, they should have increasing branches <2;3>, <4;5> ...

[tadeubas]

I only adapted the tests that already existed. I don't want to alter those, so I will create new test cases to contemplate this, ok?

[odudex]

Yes, if possible, please leave previous tests untouched, their keys and order; only change their outcomes if this was a result of the code changes. To test new scenarios, please add new descriptors, trying to reproduce usual exports of real coordinators.

[tadeubas]

Done. Plz note that this branch fixes pre-existing validation only. There are unimplemented things in Descriptor class (See TODO here https://github.com/diybitcoinhardware/embit/blob/master/src/embit/descriptor/descriptor.py#L22 ), so I had to let some TODO in tests also.

Please, fell free to submit a PR to my branch if you want any other change and I will merge right away.

[odudex]

I still see descriptors with repeated keys in tests (while I don't see in #76).
Can you elaborate on the commented TODO TR tests? What's needed for them to be activated, how can I recreate this descriptor with <0;1;3> branches (line 208) on Liana, and what's the use case for a key like this? How about others with keys that doesn't have any branches. (line 215, 222).

[tadeubas]

I'll revert the old tests, only updating the outcome check since validation now disallows miniscripts with different num_branches.

Will update TODO to note the tests don't check tr miniscript.

These descriptors are mostly edge cases to test validation, but real-world examples are included too.

[odudex]

Thank you!