refactor(fragment_cache): reuse result

[SukkaW]

What does it do?

Fix a potential performance issue introduced by #3744: After #3744, the function passed to fragment_cache might be called twice instead of once.

A PoC can be viewed here: https://runkit.com/sukkaw/5df9e053e7c3d50019810e89

const cache = {};
let called1 = 0;
let called2 = 0;

// This PR
const cacheFunc1 = (k, fn) => {
  if (cache[k] != null) return cache[k];
  
  const result = fn();
  cache[k] = result;
  
  return result;
}

// PR #3744
const cacheFunc2 = (k, fn) => {
  if (cache[k] != null) return cache[k];

  cache[k] = fn();
  const result = fn();
  return result;
}

cacheFunc1('1', () => { return called1++; });
cacheFunc1('1', () => { return called1++; });
cacheFunc1('1', () => { return called1++; });
cacheFunc2('2', () => { return called2++; });
cacheFunc2('2', () => { return called2++; });
cacheFunc2('2', () => { return called2++; });
console.log(called1, called2);
// 1 2

How to test

git clone -b fragment_cache_resue https://github.com/sukkaw/hexo.git
cd hexo
npm install
npm test

Screenshots

Pull request tasks

• Add test cases for the changes.
• Passed the CI test.

[coveralls]

Coverage remained the same at 97.117% when pulling 3a761d7 on SukkaW:fragment_cache_reuse into 5179a6a on hexojs:master.

[curbengh]

In light of this, I'm re-visiting #3744.

[SukkaW]

@curbengh I have found some similar cases introduced in #3744 as well. Should I include them in this PR?

[curbengh]

I just finished working on other cases introduced by #3744, in #3986