Manually inline function calls in a tight loop as Go compiler does not
inline non-leaf functions. It is observed that this unpleasant code
change resulted up to 10% performance improvement.

This reverts commit 376a76d1 as it
affects normal completion

Close #990

Close #1000

Running uname -sm yields:
MINGW32_NT-6.2 i686

Close #995

Removed outdated animated GIF.

https://github.com/junegunn/fzf.vim/issues/396#issuecomment-317214036

One can override the setting on FileType fzf autocmd.

I was too quick to release 0.16.9, this commit makes --ansi processing
even faster.

When --with-nth is used, fzf used to preprocess each line and store the
result as rune array, which was wasteful if the line only contains ascii
characters.

- Fix incorrect display of the last line when more than a line is
  wrapped above
- Avoid unnecessary flickering of the window

This change gives 5x speed improvement

Related: https://gist.github.com/junegunn/4963bab6ace453f7f529d2d0e01b1d85

Close #974

Fix https://github.com/junegunn/fzf.vim/issues/404

Thanks to @janlazo.

By not storing item index twice, we can cut down the size of Result
struct and now it makes more sense to store and pass Results by values.
Benchmarks show no degradation of performance by additional pointer
indirection for looking up index.

- Remove multiple mutex locks in partial cache lookup
- Simplify return values

Close #966

Related: #960 (relative filepaths)

Related #960, #552

With zero result.

Related: https://github.com/junegunn/fzf.vim/issues/22#issuecomment-311869805