For a node, maybe. What about for data processing? Say you just want to quickly find all scripts with the same destination address.
Yes there are cases in which you want to search in a script, but they are not very common. Usually you want to optimize for the most common case first.
The performance difference will be negligible for the amount of flexibility gained. Returning data like this is what RDBMS' are optimized for.
Maybe; I wouldn't be so quick to call it 'negligible', as I do know that for the current client, database accesses are by far the biggest source of performance issues.
Anyway, don't do too much premature optimisation, it might not be useful yet to discuss in this phase.