Write a test for missing tokens

[Carreau]

Missing tokens definitions when converting to Pdf

[mpacer]

Adding link: #110

[mpacer]

So in looking into this, it's easy enough to use the code linked in #110 to create an example for the purposes of testing for these particular tokens, but that seems to test for a bandaid that will only be ripped off when the token list is updated next. The changing token list has already been a problem for LaTeX template.

Looking at the preamble to the LaTeX base template, it seems that the set of tokens has changed during the course of nbconvert's development. Some of them e.g., \BuiltInTok is defined on both lines 74 and 104 in https://github.com/jupyter/nbconvert/blob/master/nbconvert/templates/latex/base.tplx (in fact, given that they both were declared with \newcommand and not \renewcommand I'm surprised it didn't throw a warning).

I've hunted down the source of the token names to be the highlighting-kate syntax highlighting engine either through pygments or pandoc. I have a hunch it's pandoc, but @jhamrick suggested it might be happening via pygments and I haven't figured out which yet… but in either case, I think we should be able to extract a list of all standard token types(regardless of the chosen style) on the fly.

And that leads my to my current question. The easiest thing to do is just run nbconvert on a notebook that happens to contain those token types, for now. The more robust solution would be to generate valid LaTeX that surely contains all the token types that can be generated by nbconvert (after extracting them directly from highlighting-kate), and ensure that that compiles.

I feel like for a test, the robust solution is preferable, but I'm not sure if that fits the goal of this issue.

[Carreau]

Simply a notebook that fail conversion would be fine for now, which should not be too hard as the conversion is just failing to PDF. Though installing latex might be too heavy.

Go for the easier to do.