Fix PoV overestimation when on-runtime-upgrade is run with checks

[liamaharon]

Closes #28

[pmikolajczyk41]

thanks for fixing this! code looks good

if I understand correctly, state_machine_call_with_proof takes only immutable references/clones, so the second simulation won't be influenced in any way by the first run

btw, there's this old comment: we don't really need any extensions here... is that true? what about situations where a storage migration uses extension for its logic?

[liamaharon]

if I understand correctly, state_machine_call_with_proof takes only immutable references/clones, so the second simulation won't be influenced in any way by the first run

correct

btw, there's this old comment: we don't really need any extensions here... is that true? what about situations where a storage migration uses extension for its logic?

I'll look into extensions and open an issue to support them if there's a use case. in the mean time, let me know if there're some concrete examples you're thinking