I think you have a number of issue combined in this question.
But about the piano sample using only middle C, you could do this but when the single sample is pitch shifted through the full piano range it will not have the right timbre.
Really good sample libraries use sample from the full range of an instrument as well as different dynamics. Really good sampling gets articulations too. Like for a violin different bowing techniques.
On the general question, can a whole song be all samples? In part I think that happens with recordings now. Drums, guitar, bass patterns can all be samples. Some parts will be from performance like vocal, or synth. Perhaps there aren't a lot of records that are 100% sampled, but I don't think there is a technical limitation. Voice is the obvious problem if you wanted to have actual text sung rather than just 'oooh' and 'aaah.'
I'm not sure if this is part of your question, but using samples doesn't necessarily mean using sequencers. I set up an old midi keyboard into a DAW, found some excellent soundfonts for various instruments and then recorded performances using those sampled soundfonts. I didn't sequence the parts.