CT descriptor: remove generic from blinding key

[LeoComandini]

The generic was only used by the Bare variant, which was inconsistent with its corresponding "secret" variant, View.

The generic could be used to handle the conversion from ConfidentialDescriptor<DescriptorPublicKey> to Descriptor<DefiniteDescriptorKey>, specifically making sure that the descriptor blinding key does not have wildcards.

For the Bare variant this could be done, but this cannot happen for the View variant, since we don't have the definite version of DescriptorSecretKey.

The proposed solution consists in:

• Remove generic from descriptor blinding key (this commit)
• Add ConfidentialDescriptor::at_derivation_index that "removes" wildcards from both the "bitcoin" descriptor and the descriptor blinding key. This makes deriving an address from a CT descriptor with wildcards less cumbersome.
• Change Key::to_public_key(&self, secp, spk) to return a Result and error if Key has some wildcards. This will fix the case ct(xprv/*,desc) which is currently broken.
• Add test cases for blinding keys with wildcards
• Add test vectors with blinding keys with wildcards to ELIP-150

[apoelstra]

Yeah, this solution seems good to me. It's not great but nothing I could come up with was great either.

I'd like to fix CI though before moving forward on this.

[LeoComandini]

fix CI

Fixed by @RCasatta in #60 , I'll rebase on top of that once merged

[LeoComandini]

Proposed additional test vectors for ELIP-150 (will post there once this gets a round of review)

View Descriptor with wildcard: ct(xprv9s21ZrQH143K28NgQ7bHCF61hy9VzwquBZvpzTwXLsbmQLRJ6iV9k2hUBRt5qzmBaSpeMj5LdcsHaXJvM7iFEivPryRcL8irN7Na9p65UUb/*,elwpkh(xpub661MyMwAqRbcEcT9W98HZP2kFzyzQQZkYnrRnrM8uD8kH8kSeFoQHq1x2iihLgC6PXGy5LrjCL66uSNhJ8pwjfx2rMUTLWuRMns2EG9xnjs/*))#

• Index: 1,
• Descriptor blinding private key: xprv9s21ZrQH143K28NgQ7bHCF61hy9VzwquBZvpzTwXLsbmQLRJ6iV9k2hUBRt5qzmBaSpeMj5LdcsHaXJvM7iFEivPryRcL8irN7Na9p65UUb/*
• Confidential address: el1qqf6690fpw2y00hv5a84zsydjgztg2089d5xnll4k4cstzn63uvgudd907qpvlvvwd5ym9gx7j0v46elf23kfxunucm6ejjyk0
• Unconfidential address: ert1qkjhlqqk0kx8x6zdj5r0f8k2avl54gmyn7qjk2k

Non-View Descriptor with wildcard: ct(xpub661MyMwAqRbcEcT9W98HZP2kFzyzQQZkYnrRnrM8uD8kH8kSeFoQHq1x2iihLgC6PXGy5LrjCL66uSNhJ8pwjfx2rMUTLWuRMns2EG9xnjs/*,elwpkh(xpub661MyMwAqRbcEcT9W98HZP2kFzyzQQZkYnrRnrM8uD8kH8kSeFoQHq1x2iihLgC6PXGy5LrjCL66uSNhJ8pwjfx2rMUTLWuRMns2EG9xnjs/*))#

• Index: 1,
• Descriptor blinding public key: xpub661MyMwAqRbcEcT9W98HZP2kFzyzQQZkYnrRnrM8uD8kH8kSeFoQHq1x2iihLgC6PXGy5LrjCL66uSNhJ8pwjfx2rMUTLWuRMns2EG9xnjs/*
• Confidential address: el1qqf6690fpw2y00hv5a84zsydjgztg2089d5xnll4k4cstzn63uvgudd907qpvlvvwd5ym9gx7j0v46elf23kfxunucm6ejjyk0
• Unconfidential address: ert1qkjhlqqk0kx8x6zdj5r0f8k2avl54gmyn7qjk2k

[LeoComandini]

fix CI

Fixed by @RCasatta in #60 , I'll rebase on top of that once merged

@apoelstra rebased and CI is now green

[apoelstra]

Looks good to me. Obviously there are still some open API questions about doing derivations but the result of this is much better than the existing code.