Add PSBT_IN_TAP_KEY_SIG field to PSBTs

[odudex]

Following guidance from @pythcoiner , we have implemented support for the PSBT_IN_TAP_KEY_SIG field. This addition addresses previously anticipated requirements, as indicated in the related comments, ensuring compatibility with internal key-signed Taproot PSBTs.

[pythcoiner]

in which case this can happen?
maybe we can avoid error here if self.taproot_key_sig == v in this case anyways?

[odudex]

I'm not exactly sure in what context this function is called, as iirc we neither use it directly in Krux nor indirectly. I simply replicated the verification from final scriptsig and final script witness, believing it is a sanity check.

[pythcoiner]

Anyway, I guess if we sign twice an input, it shold be with 2 differents nonces? (so we should end up with a different signature)

[odudex]

A question also raised by @jdlcdl: With this PR, Embit now includes the same information in both the PSBT_IN_TAP_KEY_SIG and PSBT_IN_FINAL_SCRIPTWITNESS fields. Should we retain PSBT_IN_FINAL_SCRIPTWITNESS when the signature comes from the taproot internal key? Can a coordinator wallet rely solely on the signature in PSBT_IN_FINAL_SCRIPTWITNESS, making that field still necessary?

[pythcoiner]

I dont think so, iiuc the input finalizer role must generate PSBT_IN_FINAL_SCRIPTWITNESS, and I expect the coordinator to take this role, not any signer.
But I think it's not harmfull to keep it and avoid a breaking change.

[pythcoiner]

utACK ef61cca

[jdlcdl]

utACK ef61cca

ditto!

Krux: tested and working for pr498,
SeedSigner: doesnt break existing unit-tests or screen-shot-generator, otherwise untested

untested and haven't considered how this might affect other downstream projects.

[jdlcdl]

With due respect for downstream projects, and especially for past embit contributors, tagging:

• those of who have shown an interest in taproot and/or Miniscript: @stepansnigirev , @kdmukai, @darosior, @scgbckbone, and @1ma.

• as well as others who will likely have important feedback: @achow101, @kallewoof, and @jimmysong.

Any feedback you may have for this PR, as well as for PR #67, would be greatly appreciated.

[odudex]

We plan to release a new version of Krux soon, incorporating pull requests #67 and #68. @stepansnigirev kindly granted permission to merge these changes on Embit. However, since Stepan may have other priorities, I would like to gather feedback from the other Specter and SeedSigner developers to ensure these commits are safe and beneficial for everyone.

For now, I have pushed the relevant commits to the branch psbt_in_tap_key_sig_and_reproduc_psbt (EDIT)develop. This will allow you to test the changes easily without needing to add a remote. We will continue using this branch until we reach a consensus on PRs #67 and #68, at which point I will merge them into master.

I appreciate your testing and feedback.

[odudex]

The commits from this pull request have now been merged into Embit's develop branch.

[3rdIteration]

I'll have a look at rolling a build of specter DIY with this and testing it out with Liana over the coming days