Introduce quadrature.inf to denote infinite accuracy

[inducer]

The second commit is noisy but uncontroversial. I'm mostly looking for feedback on the first.

cc @a-alveyblanc

[inducer]

Surprising! It seems like nothing in meshmode or grudge (or the tests) seems to mind this.

[alexfikl]

The first commit looks good to me!

Just wondering about the motivation. What's the reason not to just use float("inf")? Might suggest adding a little doc blurb to _Inf to justify its existence 😁

[inducer]

Just wondering about the motivation. What's the reason not to just use float("inf")? Might suggest adding a little doc blurb to _Inf to justify its existence 😁

Mostly to avoid introducing floats there as a valid data type, following the mantra "make invalid states unrepresentable". If we did, someone could claim their quadrature is exact to polynomial degree $3.5$. 🤔

I'll add a note.

[alexfikl]

Mostly to avoid introducing floats there as a valid data type, following the mantra "make invalid states unrepresentable". If we did, someone could claim their quadrature is exact to polynomial degree 3.5 . 🤔

I was thinking of something like Literal[float("inf")], but apparently that's not allowed :(
python/typing#1160