Add Integer.ceil_div function

[vrobinson-luna]

Implements Integer.ceil_div, which complements Integer.floor_div by rounding towards positive infinity when the quotient would not be an integer.

Rationale for this function

Ceiling division is quite useful for chunking up data. For instance, if you want to split data into chunks of 1024, then Integer.ceil_div(data_size, 1024) will tell you the number of chunks that will result. Or, if you want to split a list into no more than eight chunks of similar size, you can do Integer.ceil_div(length(list), 8) to get such a chunk size.

As further justification, both of these chunking use-cases happen to be present in an Elixir codebase I work in. They're currently implemented as ceil(a / b), but a pure integer version of this operation (keeping it immune to any floating-point weirdness) would be more apropos in my opinion.

As a last point, it seems enough people have asked about this in other languages to make it seem a widely useful feature beyond the situations I myself have experienced.

So if the rationale seems convincing enough for this to be in the standard library, here's a PR :)

I also expanded the unit tests for Integer.floor_div and Integer.mod to cover a few more cases with different operand signs.

[sabiwara]

The provided rationale makes sense to me, and I found occurrences of ceil(x/y) in my existing codebase at work too.
+1 for me!

[sabiwara]

(nit) the ordering of this case feels weird, would keep it after the positive args

[josevalim]

💚 💙 💜 💛 ❤️