Change Details

In D8203, the piece of the previous `AcceptToMemoryPoolWorker` function that uses `txdata` was not completely split from `PreChecks`, unlike in the source material where it was extracted to a separate`PolicyScriptChecks` function. The way this backport was done lead to `txdata` being initialized twice in Bitcoin ABC instead of just once (in Bitcoin Core or in the previous Bitcoin ABC code). This can be optimized by making `PreChecks` compute `txdata` if all the previous checks are succesfull, and pass it via an output argument to `ConsensusScriptChecks` instead of having to compute it again in `AcceptSingleTransaction`. I ran the benchmarks and found no significant difference in performance before or after this change, for any of the benchs. So this is only about cleaning the code and having the comments in `AcceptSingleTransaction` make more sense. It will also make more sense like this when we backport the second commit of [[https://github.com/bitcoin/bitcoin/pull/23381 | core#23381]]

In D8203, the piece of the previous `AcceptToMemoryPoolWorker` function that uses `txdata` was not completely split from `PreChecks`, unlike in the source material where it was extracted to a separate`PolicyScriptChecks` function. The way this backport was done lead to `txdata` being initialized twice in Bitcoin ABC instead of just once (in Bitcoin Core or in the previous Bitcoin ABC code). This can be optimized by making `PreChecks` make its `txdata` available as an output argument to `ConsensusScriptChecks` instead of having to compute it again in `AcceptSingleTransaction`. I ran the benchmarks and found no significant difference in performance before or after this change, for any of the benchs. So this is only about cleaning the code and having the comments in `AcceptSingleTransaction` make more sense. It will also make more sense when we backport the second commit of [[https://github.com/bitcoin/bitcoin/pull/23381 | core#23381]]

In D8203, the piece of the previous `AcceptToMemoryPoolWorker` function that uses `txdata` was not completely split from `PreChecks`, unlike in the source material where it was extracted to a separate`PolicyScriptChecks` function. The way this backport was done lead to `txdata` being initialized twice in Bitcoin ABC instead of just once (in Bitcoin Core or in the previous Bitcoin ABC code). This can be optimized by making `PreChecks` computemake its `txdata` if all the previous checks are succesfull, and pass it viaavailable as an output argument to `ConsensusScriptChecks` instead of having to compute it again in `AcceptSingleTransaction`. I ran the benchmarks and found no significant difference in performance before or after this change, for any of the benchs. So this is only about cleaning the code and having the comments in `AcceptSingleTransaction` make more sense. It will also make more sense like this when we backport the second commit of [[https://github.com/bitcoin/bitcoin/pull/23381 | core#23381]]