Eagerly finalize inputs

[LLFourn]

If we know the final witness/scriptsig for an input we should add it right away to the PSBT. Before, if we couldn't finalize any of them we finalized none of them. Also we don't try and re-finalize inputs that have already been finalized.

This is needed for foreign UTXOs since you don't have the descriptor for the other guy's input^.

Ready for concept acks but I'll write a test for it before merge.

^It would be nice if we could attach descriptors to PSBTs. I am not sure what the approach should be for validating finalized witness/scriptsig.

[afilini]

Definitely concept ACK, it seems a much smarter approach compared to what we were doing before!

[afilini]

Now that I think about it, wouldn't it make more sense to also return another boolean (or maybe, for clarity, a struct with named fields) to tell the user whether we've actually signed something? Because currently we only have is_finalized, but that's not really useful if you have inputs from multiple parties since generally you can't finalize the tx entirely, but maybe you can sign a few inputs.

[LLFourn]

Test added. It's based on this PR so it can compile: #187. Ready for final review once that's sorted.

@afilini I see your point. It would be nice to know if progress was made. For my use the "is_finalized" is enough to get what I need done. I don't have a strong vision for that yet though so maybe another PR unless your confident in a particular design?

[afilini]

I don't have a strong vision for that yet though so maybe another PR unless your confident in a particular design?

Yeah that's good for me, I'm still not sure if it's better to just say "i've signed something" or maybe something more detailed like a list of which keys signed, etc. So for now it can stay like that and we can have a future upgrade to add those stuff, maybe as part of a broader refactoring of the whole policy module which really needs some work.