They are not as separate. A deferred shading pipeline would make little sense if it was operating purely in LDR. Tonemapping in foward shading can be done but makes a lot more sense to do as a post step. If you ask a photographer about tone mapping he/she would tell you about a very different concept (but is more likely to turn away in disgust)... and that could not be done by color-mapping 1 fragment at a time.