We should follow a philo of "do not do the same computation twice", thus, any time we minify or compress a resource, we should cache this to be able to re-serve it MUCH faster with zero computational overhead or wasting of CPU cycles.

This should be optimised in a way that tries to balance between memory preservation, and maximise "hot" / frequent file requests hitting the cache, infrequent files or noticable traffic drops should prompt a drop from the cache, somewhat like a Garbage Collector.

We need to handle missing files properly (return 404s), ensure full integrity and stability of all the internal transformations (minifier, brotli, ut8<--->byte conversion ops, file parsing, etc), we cannot let any errors go uncaught and especially not let them leak into responses.

If the request itself causes issues, such as missing files, weird request input, then we toss back the expected HTTP code (bad request).

The daemon uses hardcoded dummy values, we need to allow total customisation of the daemon... from network interface, to minifier and brotli parameters, the ability to determine which directory to host from, and what 'base route' to serve from, there's plenty of options we should allow.

Currently the content-type is hardcoded, which is obviously stupid, we should detect the content type from the file extension and pass it correctly to the response headers.

Simple task, but we need to find a way to map 'extension' to 'content-type', perhaps a crate for this?