Use unchecked to improve gas usage of ERC721

[Amxx]

Following the logic in described in https://github.com/OpenZeppelin/openzeppelin-contracts/issues/3512#issuecomment-1172614573.

• _balances are private
• _balances are only increased/decreased when tokens are minted, burned, or transferred

A balance cannot decrease below 0 because that would require transferring/burning more tokens out of a wallet than were transferred/minted to this wallet in the first place.

The balance cannot overflow, because that would require minting all the tokenIds, and putting them all in the same wallet (there are 2**256 tokenIds, so that would overflow by one). While this is mathematically possible, it is impossible in practice (with token being transferred/minted one at a time)

PR Checklist

• [ ] Tests
• [ ] Documentation
• [ ] Changelog entry

[frangio]

Given that we want to explore batch minting in ERC721 it seems we should not merge this, as the assumption that minting happens one at a time will be invalidated. Also please include benchmarks, what kind of improvement do you see with this PR?

Use of unchecked should be accompanied by comments explaining why it's safe.

[Amxx]

Benchmark of the changes (running the transactions in test/token/ERC721/ERC721.test.js)

[Amxx]

I understand that we want to experiment new minting mechanism. However, I believe that these minting mechanisms should remain ERC721 compliant, which means emitting a Transfer event for all the tokens ids. This means that, in practice, it's impossible to mint all the tokens, and the balance increment overflow is not an issue.

[Amxx]

For reference: https://twitter.com/optimizoor/status/1544626818996613120?t=8tV84gD3blTSsxk14EotiQ

TLDR: batch minting should be limited to "not to big batches" ... which means that the limit should never be reached, even minting 10**18 token each time

[Amxx]

I believe once #3611 is merged, the requested changes will no longer be needed.