Month of testing and profiling show that fasthttp doesn't gives us significatnt profit in memory and performance while being incompatible with many third-side packages and http/2
* Add ETag caching support
* Update the readme file
* Typo
* Calculate ETag before image processing, rename Env Var
* Make PO struct field public
* Generate random value on server startup to be used in ETag calculation
* Minor refactoring
* Adjust boolean value parsing as discussed in GitHub PR page
* Move random value generation to the config
* Update README
* Revert changes
* Use footprint to calculate the hash of an image