Gas optimize ERC5643

[ByteEVM]

I have applied some optimizations on repository-owned code whilst keeping it readable:

Changes

• Bump compiler version from 0.8.13 -> 0.8.19, right before PUSH0 opcode was introduced, thus allowing the usage of this base on not fully EVM compliant chains.
• Replace solidity revert errors with their inline-assembly equivalence:
  • Declarations of errors were not removed to support recognition by other libraries.
• Emit SubscriptionUpdate logs and perform storage-related operations with inline-assembly.
• Perform other casting and comparisons with inline-assembly to further optimize checks.

All forge tests passed successfully, with these before and after gas costs:

• testCancelNotOwner(): from 17,836 to 17,800
• testCancelValid(): from 20,777 to 20,723
• testExpiresAt(): from 42,743 to 42,056
• testExpiresAtInvalidToken(): from 12,901 to 12,865
• testExtendSubscriptionInvalidToken(): from 13,039 to 12,991
• testIsRenewableInvalidToken(): from 12,831 to 12,795
• testRenewalDurationTooLong(): from 47,961 to 47,581
• testRenewalDurationTooShort(): from 47,664 to 47,583
• testRenewalExistingSubscription(): from 145,175 to 144,285
• testRenewalInsufficientPayment(): from 26,874 to 26,650
• testRenewalInvalidTokenId(): from 22,316 to 22,304
• testRenewalNewSubscription(): from 53,103 to 52,702
• testRenewalNotOwner(): from 17,961 to 17,913

Notes

• More optimizations would have been possible (e.g. using STATICCALL) at the cost of code readability.
• Further gas optimization suggestions would be relying on gas-efficient libraries such as ERC721A for NFT allowance and existence checks.