Reducing the deployment size of the NFT descriptor with zipped contracts

[PaulRBerg]

@razgraf suggested taking a look at zipped-contracts for potentially reducing the deployment size of our NFTDescriptor by up to 50%.

After skimming through the README and the launch thread, these are my thoughts and remarks particularly relevant to Sablier:

• It is an interesting design, but it relies upon a whole bunch of inline assembly, and a quasi-experimental Solidity implementation of the zlip deflate algorithm.
• Zipped contracts cannot be interacted with via Etherscan. For this, we would have to write a separate wrapper that calls the zip.
• Certain features, e.g. address(this), are not compatible with the zipped design. This is not great for us given our protocol's dependency on many runtime features like block.timestamp.

Given the points above, I suggest waiting before using this zipped contract design, i.e., we should let other (simpler) projects pick it up and make it more robust. Also, it isn't audited at the moment.