Enforce a strict number type across the code.

[maschad]

Throughout the repo we use u64 types as well as i64 types, I suppose the latter is due to postgres using i64 to represent their BIGINT but this of course can lead to loss of precision, particularly when converting from floating point types as well. So we need a consolidated approach to handling these types.

My recommendation would be to standardize signed integer types across the entire repo.

[jorgeantonio21]

I might be wrong, but I don't recall any cast from float points to i64 directly. Any concrete example ?

[maschad]

@jorgeantonio21 you are correct, forgive my poor wording, I should have said "converting to floating point types"

But we actually do cast from i64 to f64 such as here for instance, and casting i64 to f64 causes a loss of precision (i64 is 64 bits wide, but f64's mantissa is only 52 bits wide)

[jorgeantonio21]

I see, that said I believe these cases are pretty safe to cast, but we should definitely flag them all