Then the usage example athttps://matplotlib.org/api/type1font.html#module-matplotlib.type1font is misleading because it's not actual a use case? It explicitly creates single-item tuples and passes them to the function.

Sorry, I guess my comment was slightly misleading. The entries in pdftex.map don't necessarily specify both sland and extend, the idea is more that the two of them (or only one of them, if only one of them exists) are carried together as a dict through multiple functions and it doesn't make sense to split them at the last minute.

Will decide later if I want to follow up on this.

at least the (independent) docstring cleanups are fine by themselves.

(I guess one could introduce a namedtuple to represent the slant/extend pair, but that feels a bit overengineered to me -- there's some level at which dicts are just fine...)

Not going to fight over this.