The idea here is to parse the typical mime.types
format used by web servers like
Apache HTTPD to infer Content-Type values based on filename suffixes, and then
convert the result into the map structure expected by the shared module
hashicorp/dir/template
.
This can therefore get a similar effect to serving files from a static web server,
using the same mime.types
as was used on that server in order to ensure selection
of the same content types.