Optimize arity wrapper for << and >>

[jfmengels]

Prior to this, << (and similarly >>) would be compiled to

var $elm$core$Basics$composeL = F3(function (g, f, x) { ... })

which doesn't match the call sites, which is always of the shape

A2($elm$core$Basics$composeL, f, g)

This change makes it so that composeL/composeR are now wrapped with A2, which improves the performance by about 30%.

---

Here is the benchmark that I have been using: https://github.com/jfmengels/elm-benchmarks/blob/master/src/WhatIsFaster/FunctionComposition.elm (where I compare >> and lambdas)

Prior to the change:

After the change:

---

A further improvement to this (not in this PR, and more in the scope of the compiler would be to replacing A2(composeL, f, g) by composeL(f, g):

---

PS: At some point in the past, the Elm compiler turned composition into lambdas, but it removed that in elm/compiler@6ae0168.

[github-actions]

Thanks for suggesting these code changes. To set expectations:

• Pull requests are reviewed in batches, so it can take some time to get a response.
• Smaller pull requests are easier to review. To fix nine typos, nine specific issues will always go faster than one big one. Learn why here.
• Reviewers may not know as much as you about certain situations, so add links to supporting evidence for important claims, especially regarding standards for CSS, HTTP, URI, etc.

Finally, please be patient with the core team. They are trying their best with limited resources.