Naming of reference shape in docs

[KnutAM]

As discovered in #935, we currently denote AbstractRefShape in the docs as

• reference shapes
• reference cells
• reference elements

I would suggest to make this consistent throughout the docs in a pure rename PR, and since the type is AbstractRefShape suggest using the first name consistently.

[termi-official]

Related to this, so I answer here. Maybe we should add in the nomenclature section (xref #944) and maybe in the beginning the relation between the wordings, as newcomers might wonder about why we do not call everything Element in the finite element method.