Would it make sense to pre-allocate this `HashSet` with a capacity of the sum of `input_native_tokens.len()` and `output_native_tokens.len()` so this doesn't ever have to re-allocate?

[thibault-martinez]

Would it make sense to pre-allocate this HashSet with a capacity of the sum of input_native_tokens.len() and output_native_tokens.len() so this doesn't ever have to re-allocate?

Originally posted by @Alex6323 in https://github.com/iotaledger/bee/pull/1331#discussion_r843020216