refactor: remove ramda from layout package

[diegomura]

Gradually start removing ramda. Although I like it, it's probably confusing for most people, it's slow and adds unneccesary bundle size. Some helper fns will still be needed. For the moment I moved them to utils dir, in the future it worths building our own set of light and reusable utils

[jeetiss]

I made a benchmark and the result is insane, almost 20% speed up on a simple document with one page.

master:

yarn run v1.22.17
$ babel-node ./bench/index.jsx

Running "simple" suite...
generate doc x 33 ops/sec ±7.7% (60 runs sampled)
✨  Done in 8.22s.

your branch:

yarn run v1.22.17
$ babel-node ./bench/index.jsx

Running "simple" suite...
generate doc x 40 ops/sec ±8.9% (63 runs sampled)
✨  Done in 7.80s.

[diegomura]

Wow that's great! Thanks for checking @jeetiss !

In addition to make code more vanilla that's the main motivation behind this effort, glad the benefits are visible. Using Object.assign instead of spread and small changes like that are aligned to this as well.

And that's not even considering ramda still dominates the textkit, stylesheet and render packages (textkit gains specifically should be big, considering text layout is one of the most expensive pieces here). So expect more improvements!

[diegomura]

I should be done removing ramda everywhere in some few days more of work. There's an open PR to remove it completely from @react-pdf/stylesheet. Still to go some cases in @react-pdf/layout (this PR), @react-pdf/textkit (PR in progress, by far where we use it the most, which sounds promising), and @react-pdf/render (no PR nor progress yet)

[diegomura]

@jeetiss with upstream changes now benchmarks look even better in this branch 😄

And there are still packages to migrate. It's a tedious work but most if it it's done already

generate doc x 89 ops/sec ±4.84% (76 runs sampled)
✨  Done in 8.57s.