Resolver performance optimizations + optional local caching for better re-run performance

[vickygos]

Goal

• To Enable quicker iteration of OpenRarity scoring algorithm changes
• To enable quicker individual runs of resolver

Changes

• Added in optional caching whenever we fetch data from external API's (opensea for trait data, and external ranking providers) by storing the fetched data in local json files used for reading in subsequent runs. This allows re-runs to be lightning fast as they do not require re-fetching of data through API's as they tend to be static, especially in the short-term.
  • It also means that users with trait or ranking data in their own databases can use that by following the same json cache format, thus avoiding API calls.
• Optimized resolver so that we bulk fetch all opensea tokens prior to fetching external resolver ranking data.
  • This allows us to have higher batch limits for external resolvers (like traitsniper, batch = 300), even though opensea API limits are 30.
• Sample BAYC trait data json cache file to memoize opensea API calls.

Unlocks

• If we are testing various algorithm changes, and want to get the diffs csv from resolver, we can use the default cache field to avoid external API calls in subsequent runs.

[theelderbeever]

As a pattern maybe make a .cache directory and then just gitignore .cache? That way it isn't file type dependent and doesn't show up in the repo?