Add caching to aggregator responses

[vlad-ignatov]

There are multiple caching strategies we can try. I'd propose we start with the simplest, which could look like this:

1. You get the modification date of the underlying s3 file
2. Make sure you send that date in a Last-Modified header
3. You may also get a if-modified-since header. If so, compare and:
   • Last-Modified > if-modified-since - exit with 304 status code without doing anything else.
   • Last-Modified <= if-modified-since - Continue with the queries as usual

Notes:

• If you send Last-Modified you should get back if-modified-since in the follow up requests. Otherwise it may not be present.
• The actual storage in this case happens in browsers.
• Browsers have mechanisms to clear the cache or to bypass it by force reloading (like Cmd+Shift+R on Mac)
• If this is not good enough (if the file modification timestamp is not the only variable we need to account for) we can switch to ETag caching which is very similar
• See https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Last-Modified and https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/If-Modified-Since