Removal of curry2 and curry2flipped

[eborden]

These methods become redundant with the introduction of curry and rCurry.

[raganwald]

My allong.es library has both curry2 and generic curry functions because it does a lot of binary currying itself, and curry2 is faster than curry. But I can see the argument for keeping the API simpler.

[fogus]

I think a case can be made for fast versions of 2 and 3-arg currying. I can't personally recall a time when I've curried more than 3. I'd be open to putting curry2, rcurry2, curry3 and rcurry3 (back) in. Maybe another ticket is in order?

[eborden]

I'd be curious to see a perf of these functions. I'm sure curry2 is more performant, but the proof is in the pudding

[fogus]

Moved discussion to #42