Protocol/Michelson: UNPAIR instruction
This MR adds an
UNPAIR instruction that behaves as the current
Motivations for doing it:
UNPAIRmacro is very widely used, promoting it as an instruction diminishes its gas cost which profits almost all contracts (this was the purpose of the incorrect implementation of the peephole optimisation that Carthage removes)
- providing an alternative to the
CDRinstructions seems required for linear typing; a feature that might be desirable for value stamping (see https://forum.tezosagora.org/t/contract-signatures/1458). Note however that linear alternatives to
UPDATEshould also be added.
Because of the ubiquitous use of the
UNPAIR macro in the testsuite, the difference in the regression log is unfortunately too big to be checked by a human reviewer.