Golfing ERC721A

[0xPhaze]

Hey,

I tried bringing the costs down for ERC721A. Here's a gas report with a small readme:
https://github.com/0xPhaze/ERC721A-Improved

• Internally, tokenData is passed as storage instead of memory.
• Keeps track of whether the next token data has already been set before, so that we don't have to check it (1 cold sload) on every transfer for all eternity.
• Saves (warm) sload on variables such as collectionSize and maxPerWallet.

Function	Gas
mint 1
test_mint1_ERC721A()	59713
test_mint1_ERC721AX()	59333
mint 5
test_mint5_ERC721A()	67486
test_mint5_ERC721AX()	67131
transfer 1
test_transferFrom1_ERC721A()	73837
test_transferFrom1_ERC721AX()	49435
transfer 5
test_transferFrom5_ERC721A()	154195
test_transferFrom5_ERC721AX()	127657

Wanted to ask first, if ERC721A is interested having some of these optimizations implemented.

[Vectorized]

@0xPhaze I've added in #272. Will credit you in the next release. Big thanks!

I've opened a PR to your repo explaining the issue with using storage modifer in your case.

[0xPhaze]

Awesome, thanks!
Although, I think that doesn't include the 'nextTokenDataSet' boolean which is a big part of the optimization. Instead of checking whether the next packed ownership has been set already by loading its storage, this would keep track of whether it has already previously been set and save that sload for all future transfers. And it can directly be assumed to be set (meaning we never have to check) this if the user only minted one.