Lucene needs text to index something. Search indexing relies on ability to convert a specific format to text - if it can convert it to text, it will index it. If we could convert images to text, then we could index them…. So, if you configure your transformations in a way that .tmp file cannot be converted to text, then the indexing is effectively disabled.
I don't believe there are other provisions for turning off indexing for certain mime types.
