Replace num_bigint with primitive_types::U256?

### What is wrong?

From #410:

We may need to reconsider the usage of BigInt,
I think it would be better to use U256 because:

1. it provides more reliability than multiple-precision arithmetic.
2. it improves performance a little bit.

### How can it be fixed

@Y-Nak? I don't really have any feelings about this; just copied it here so it doesn't get lost in the closed PR.

Start by reviewing the current BigInt usage and the context in issue #410, then compare it with primitive_types::U256. Determine the full migration scope and compatibility implications before changing anything; done means the project consistently uses U256 where intended and its existing behavior remains covered.

Written by the indexing model from the issue text.