Add reject and rejectM

[svozza]

Add inverse versions of filter and filterM.

[gabejohnson]

I'm not sure what you mean by inverse versions. What is the expected behavior?

[davidchambers]

I believe the issue title should mention reject and rejectM, but I'll let Stefano confirm this.

[svozza]

Yes, that's it. Apologies, I should have been explicit.

[gabejohnson]

Is this necessary since we have complement?

[davidchambers]

These functions aren't necessary, but they do improve readability in some cases. Consider:

> filter(complement(negative), [-2, -1, 0, 1, 2])
[0, 1, 2]

> reject(negative, [-2, -1, 0, 1, 2])
[0, 1, 2]

I find myself exerting significantly less effort when reading the second expression.

[gabejohnson]

Just playing 😈's advocate.

[svozza]

Yeah, I find complementstoo clunky.

[davidchambers]

It's worth noting that complement is itself unnecessary. This is the desugared form:

> filter(map(not, negative), [-2, -1, 0, 1, 2])
[0, 1, 2]

Although the (not, negative) section actually reads quite nicely, having to reason about lifting not makes this much less approachable than the reject equivalent.

Put differently, complement is nice sugar and reject is even nicer sugar in such cases.