additional information for `finalize_psbt` when returning false

[reez]

Describe the enhancement
I'm wondering if anyone has potentially thought of any alternatives to the return type than what currently exists for sign Result<bool, SignerError>

Update: I'm scoping this issue down to Wallet::finalize_psbt rather than Wallet::sign. The concrete enhancement is to expose more detailed finalization results than a single boolean, so callers can inspect what happened for each input when finalization is incomplete.

Use case
I'm specifically thinking about a client app signing a transaction and getting bool of false back without additional information of why sign returned false.

Maybe they were expecting it to return false but maybe not; that client app may expect successful signing, not receive true or SignerError back, but get back false instead without extra context, on what could be considered an imporant operation for their app.

Update: For finalize_psbt, this means a caller can currently receive Ok(false) without knowing whether a specific input was already finalized, finalized during the call, missing a matching descriptor, or could not be satisfied. Returning per-input finalization outcomes would give client applications enough context to decide what to surface to users or how to continue processing the PSBT.

Additional context
Just looking for feedback/thoughts if there has been conversation about this historically/presently, and any possible thoughts.

Update: Wallet::sign return-type changes are now out of scope for this issue and can be handled separately as part of the broader signing API transition. #433 is in progress for the finalize_psbt side of this issue and should close this if merged.

[A-y-ush]

Hey, so from what I understand . The problem we are trying to solve is explaining why the transaction has not been signed (completely) which means the signing is done but the PSBT is not finalised yet. If I am understanding it right, it does make sense to add the potential reasons for this. Although I am also unaware about any prior discussion on this.
we can create separate cases for which the PSBT is not finalised during the transaction and return those.

[A-y-ush]

@reez can I get assigned to this issue? I would like to work on it.

[reez]

@A-y-ush I don't seem to have access on this specific repo to assign this to anyone, I think this is a good item for me to mention on this weeks Lib Dev Team Call to check in on other folks thoughts on how they'd like to proceed as well as get it assigned to you, thanks for the feedback and volunteering to take this on!

[reez]

@A-y-ush you've been assigned now, this is a good research topic, and may not make it into the final beta release or anything, but would love to see what you come up with as a possible solution, thanks!

[ValuedMammal]

Recall that if try_finalize is set, then sign just propagates the result of finalize_psbt, so the issue should be addressed there. I would follow a similar approach as FinalizeMap and return a structure containing the result of finalization at each input index. Finalization can fail if either the descriptor can't be found or a satisfaction can't be created for a signed input. The method should only throw an error due to a malformed PSBT (e.g. input index out of bounds).

Instead of a boolean, sign could return Ok(()) indicating signing was successful, or better would be return Ok(SigningKeysMap) the way Psbt::sign does.