Change doubly LL to singly LL in hash table

[iLeafy11]

No description provided.

[jserv]

Run clang-format -i *.[ch] for consistent style.

[jserv]

Update .clang-format to recognize hlist_for_each as a for-each macro.

[iLeafy11]

Done!

[jserv]

You shall indent this source file accordingly.

[jserv]

Read https://cbea.ms/git-commit/ carefully and refine your git commit messages.

[jserv]

Improve git commit messages. You should summarize the motivation and consideration to make such changes. In addition, the subject ought to be revised.

[iLeafy11]

There'll be errors occur when having an input like this:
a aaa a
Will refine the git commit messages after solving this issue.(solved)

[jserv]

There'll be errors occur when having an input like this: a aaa a Will refine the git commit messages after solving this issue.(solved)

Would you continue working on this pull request?

[iLeafy11]

There'll be errors occur when having an input like this: a aaa a Will refine the git commit messages after solving this issue.(solved)

Would you continue working on this pull request?

Yes, I’ll continue to work on it.

[jserv]

Yes, I’ll continue to work on it.

@iLeafy11, I look forward to your work. Feel free to submit further commits.

[iLeafy11]

According to perf experiment on word-count implementations using different data structures:

Execution time comparison (32 runs):

• word-count-hlist: ~11.5 ms
• word-count-list: ~10.2 ms

The hlist version having slightly worse execution time, so I took a deeper look into cache behavior:

Cache performance:

• hlist version: 1457801 cache refs, 13.57% cache misses
• list version: 1484383 cache refs, 20.85% cache misses

This indicates that the hlist version has a cache performance advantage.

However, its performance is likely hurt by branching overhead introduced in ht_insert():

if (!tail)
    hlist_add_head(...);
else
    hlist_add_behind(...);

Therefore, I'm unsure whether this direction is worth continuing.