Revisit SessionHistory::psbt_with_fee_contributions

[spacebear21]

psbt_with_contributed_inputs still seems like a hack since the caller has the inputs to manage itself, and Liana should probably handle liana-specific details, but no reason for a veto. maybe the name should actually be unsigned_proposal_psbt. We can revisit this when reviewing the liana integration

Originally posted by @DanGould in #750 (comment)

[arminsabouri]

These methods should have been pruned in the original PR #750

[spacebear21]

Ah my bad I missed that. Updating this issue for a specific remaining follow-up from the same comment thread.

[spacebear21]

psbt_with_contributed_inputs was renamed to psbt_with_fee_contributions in 3dcea20.

[arminsabouri]

What would be the closing criteria for this ticket? Perhaps we need to provide rationale for why this method needs to exist? And remove if we don't need it?

[spacebear21]

That sounds appropriate.

[arminsabouri]

Just took a look at the Liana impl. Lianad needs to save a "ready-to-be-signed" PSBT in the db after fee's have been applied so liana-gui can actually sign. When lianad is responding to the sender it also uses psbt_with_fee_contributions to look up the PSBT by txid in db and use that PSBT to respond to the sender. In both cases a pub method off ProvisionalProposal will suffice.

The demand for such a method or field really comes from the fact that the entity / part of the system that progresses the typestate is different from the entity that has access to the private keys.

[arminsabouri]

A method off the typestate is the most idiomatic approach for reading.