Another problem is that some of the functions we might want to inline (such as uncurry wrappers) do not yet exist at the time of module elaboration.
The inline predicate is the interesting bit.
Each function call carries some overhead, so it may be advantageous to inline functions, by expanding the function body at the places where the function is used.
A simple solution to this problem is to inline only those functions that are used exactly once and whose actual parameters are just atoms (variables or literals).
Only the unknown calls are included: we don't care about known functions because we generate inline code for their calls.
The implication here is that any inline images to be included need to be carefully optimised in terms of the file size.
Requests for inline images (indicated by the tag) are evaluated eagerly and displayed inline.
As an example, consider the following definitions of functions foo and bar where we wish to inline all calls of foo.
