In the spirit of using go with parallalism, the template file loader now
uses goroutines, wait groups and channels to parse/load templates.
Although the performance benefit personally is slight, it did highlight
that other users of this lib would have issues with the default loader
struct not having a sync.Mutex available. Required when parsing in
goroutines.

This also meant it was prudent to add a timeout on load so that an error
is returned if unexpected time to load. Because this package is used for
me personally in a private repo, I also had to create the go mod file so
that it could be referred to locally easily with the replace directive.

A new test was made and the readme updated with the new options.