Replace `indexOf === 0` with `startsWith`

[Juneezee]

A small PR.

It is more readable to call startsWith than to call indexOf and compare the result with zero to determine whether a string starts with a given prefix.

startsWith is also more efficient as it only compares at the beginning of the string, while indexOf searches the entire string.

[ai]

Can you run https://github.com/postcss/benchmark before and after the change?

[Juneezee]

Before:

pnpm test parsers

> @ test /home/jun/Desktop/github/benchmark
> eslint . && gulp "parsers"

[22:21:34] Using gulpfile ~/Desktop/github/benchmark/gulpfile.js
[22:21:34] Starting 'parsers'...
[22:21:34] Starting 'bootstrap'...
[22:21:34] Finished 'bootstrap' after 3.7 ms
[22:21:34] Starting '<anonymous>'...
[22:21:34] Running suite Parsers [/home/jun/Desktop/github/benchmark/parsers.js]...
[22:21:40]    Rework x 12.01 ops/sec ±1.70% (35 runs sampled)
[22:21:46]    Next PostCSS x 19.77 ops/sec ±7.35% (54 runs sampled)
[22:21:52]    PostCSS x 19.88 ops/sec ±8.45% (55 runs sampled)
[22:21:58]    PostCSS Full x 7.39 ops/sec ±7.63% (33 runs sampled)
[22:22:04]    CSSOM x 25.10 ops/sec ±7.48% (48 runs sampled)
[22:22:09]    Mensch x 15.63 ops/sec ±7.72% (46 runs sampled)
[22:22:17]    Gonzales x 3.76 ops/sec ±4.26% (14 runs sampled)
[22:22:23]    CSSTree x 15.88 ops/sec ±5.30% (45 runs sampled)
[22:22:30]    ParserLib x 3.81 ops/sec ±1.90% (14 runs sampled)
[22:22:37]    Stylecow x 8.14 ops/sec ±5.30% (25 runs sampled)
[22:22:42]    Stylis x 47.36 ops/sec ±2.29% (64 runs sampled)
[22:22:42] Fastest test is Stylis at 1.89x faster than CSSOM

Stylis:       21 ms  (2.4 times faster)
CSSOM:        40 ms  (1.3 times faster)
PostCSS:      50 ms
Next PostCSS: 51 ms  (1.0 times slower)
CSSTree:      63 ms  (1.3 times slower)
Mensch:       64 ms  (1.3 times slower)
Rework:       83 ms  (1.7 times slower)
Stylecow:     123 ms (2.4 times slower)
PostCSS Full: 135 ms (2.7 times slower)
ParserLib:    263 ms (5.2 times slower)
Gonzales:     266 ms (5.3 times slower)

[22:22:42] Finished '<anonymous>' after 1.13 min
[22:22:42] Finished 'parsers' after 1.13 min

After:

I updated package.json to use my local postcss by "postcss": "file:../postcss"

pnpm test parsers

> @ test /home/jun/Desktop/github/benchmark
> eslint . && gulp "parsers"

[22:23:42] Using gulpfile ~/Desktop/github/benchmark/gulpfile.js
[22:23:42] Starting 'parsers'...
[22:23:42] Starting 'bootstrap'...
[22:23:42] Finished 'bootstrap' after 3.69 ms
[22:23:42] Starting '<anonymous>'...
[22:23:43] Running suite Parsers [/home/jun/Desktop/github/benchmark/parsers.js]...
[22:23:49]    Rework x 12.09 ops/sec ±2.44% (35 runs sampled)
[22:23:54]    Next PostCSS x 20.58 ops/sec ±7.20% (56 runs sampled)
[22:24:00]    PostCSS x 19.58 ops/sec ±8.67% (54 runs sampled)
[22:24:07]    PostCSS Full x 7.50 ops/sec ±6.63% (34 runs sampled)
[22:24:12]    CSSOM x 25.09 ops/sec ±4.69% (47 runs sampled)
[22:24:18]    Mensch x 16.14 ops/sec ±8.19% (47 runs sampled)
[22:24:25]    Gonzales x 3.91 ops/sec ±4.09% (14 runs sampled)
[22:24:31]    CSSTree x 16.46 ops/sec ±5.26% (47 runs sampled)
[22:24:39]    ParserLib x 3.76 ops/sec ±2.57% (14 runs sampled)
[22:24:45]    Stylecow x 8.31 ops/sec ±6.19% (26 runs sampled)
[22:24:51]    Stylis x 46.37 ops/sec ±8.02% (64 runs sampled)
[22:24:51] Fastest test is Stylis at 1.85x faster than CSSOM

Stylis:       22 ms  (2.4 times faster)
CSSOM:        40 ms  (1.3 times faster)
Next PostCSS: 49 ms  (1.1 times faster)
PostCSS:      51 ms
CSSTree:      61 ms  (1.2 times slower)
Mensch:       62 ms  (1.2 times slower)
Rework:       83 ms  (1.6 times slower)
Stylecow:     120 ms (2.4 times slower)
PostCSS Full: 133 ms (2.6 times slower)
Gonzales:     256 ms (5.0 times slower)
ParserLib:    266 ms (5.2 times slower)

[22:24:51] Finished '<anonymous>' after 1.13 min
[22:24:51] Finished 'parsers' after 1.13 min

[ai]

Thanks!