perf(tiny-react) optimize reconcile

[hi-ogawa]

ideas

• can skip render when identical vnode? (though it usually effective only for memo component? maybe not)
  • cf. https://github.com/preactjs/preact/blob/19de3d9688441a227af3a9edb521905e4b3d8195/src/diff/index.js#L138-L147
• more early return for memo code path?
• move placeChild logic to be responsible of "caller" side (e.g. VFragment.children)?
• optimize jsx runtime
  • delay props destructuring to render time

Maybe 1st and 4th factors are what differentiates for 04_select1k benchmark between tiny-react and react hi-ogawa/js-framework-benchmark#1

Related to the 1st point, isn't it safe to default to apply memo automatically to all functional component (i.e. no re-render if props shallow equal)?

• https://yew.rs/docs/advanced-topics/optimizations#pure-components

[hi-ogawa]

superseded by

• perf(tiny-react): render optimization for "identical vnode" + fix memo #173
• refactor(tiny-react): handle children re-insertion in fragment reconcilation #169