I do not think that a blacklist depends on genome size. Blacklisted regions are those that (regardless of the specific protein target/antibody) always attract reads and form peaks, for example because they attract spurious mappings. That though you would only know if you have many datasets from that bacterium and could therefore identify these consistently-enriched regions. If you don't have that, then you probably cannot make a reliable blacklist. ENCODE to my knowledge has NGS blacklists for very few models (https://github.com/Boyle-Lab/Blacklist/tree/master/lists) but not any bacteria. That means you probably have to accept the peak lists you get. Others might have different opinions.