What is so bad about doing IBR on non-html content?

What is the rationale behind only doing Intelligent Browser Referencing on HTML pages and then only on certain types of links within them?

 

 

Assuming there is no supported way to expand the set of content types that are covered by IBR, what if someone kludged a way for it to happen? What type of undesirable unintended side effects could/would that bring about?

 

 

Is there not something to be gained by doing IBR on ...

 

 

css files that contain links to images?

 

 

js files whose sole purpose in life is to *include* other js files?