feat: parallel pre-processing of precompile results

[ARR4N]

Many precompiles are pure functions, which can therefore be computed in parallel if their inputs are known ahead of time. Add support for collection and dispatch of transaction pre-processing.