Theory derivations should be more than simple `Sentence`s

[niazim3]

Problem from #1356: We want to be able to place figures in theory derivations because they can aid in understanding theory derivations.

In order to do this, this original ticket proposed re-writing Derivations to rely on [Content] instead of [Sentence] (i.e., a list of arbitrary document content as opposed to [scientific] natural language). Is this the right solution? I'm not immediately sure. We should (1) figure out what things should go in derivations, (2) figure out how we actually want to build derivations, and then (3) evaluate this as a solution vs other choices.

Original description from 2018:

(Moved from #963): As mentioned in today's meeting, Derivation should be changed from https://github.com/JacquesCarette/Drasil/blob/fb0156177d785df905e0b0147b7b2b5ea57b94f9/code/drasil-lang/Language/Drasil/Chunk/Derivation.hs#L8 to type Derivation = [Contents] to eliminate the need for mkEmptyLabel here: https://github.com/JacquesCarette/Drasil/blob/fb0156177d785df905e0b0147b7b2b5ea57b94f9/code/drasil-docLang/Drasil/DocumentLanguage/Definitions.hs#L67-L76