[wip] Implement redesigned PSET

[achow101]

PSET is being redesigned in order to actually do everything that elements transactions can do while also preserving privacy.

See ElementsProject/elements#900 for more details about this, including a specification for the new PSET.

[jgriffiths]

@niftynei Heads up that the elements interface will be changing per the linked PR.

Hi @achow101, for the presence/absence of fields in inputs and outputs I am leaning towards using a bitset and removing the hacks to malloc(1) buffers leaving their length 0. Can you comment as to how likely the Elements interface is to change at this time?

[achow101]

Can you comment as to how likely the Elements interface is to change at this time?

I'm not sure how it will change at this time, there still some work being done on the design of the Blinder role.

[jgriffiths]

I'm not sure how it will change at this time, there still some work being done on the design of the Blinder role.

OK, I imagine we want the updated Elements support in the next release, since we can then get all the PSBT interface changes done within one release.

[niftynei]

Suggested change

	#define PSET_GLOBAL_SCALRA 0x00
	#define PSET_GLOBAL_SCALAR 0x00

extremely drive by nit lol

[jgriffiths]

@achow I have started implementing these changes locally including tests, swig/C++ wrappers and code de-duplication. Please see https://github.com/jgriffiths/libwally-core/tree/elements_pbst. I've fixed the malloc hack in that tree.

I'm thinking about moving all varbuff members of inputs and output to just be held in a map internally. That seems easier than fiddling around with struct members while the interface settles, and it would make serialisation easier too.

[jgriffiths]

Closing this in favour of #218 which implements the PSET BIP as currently specified. Please update there when the spec changes/is finalised.