[Fix] utils: respect encoding of surrogate pairs across chunks - #559

Summary

• avoid splitting UTF-16 surrogate pairs across the encoder's 1024-code-unit chunks
• add a regression test for a surrogate pair that lands on the chunk boundary
• rebuild the browser bundle

Bug

utils.encode chunks long strings before percent-encoding. When a high surrogate lands at the end of a chunk and the low surrogate is in the next chunk, the pair is encoded as two invalid code points instead of one Unicode character.

Example before this change:

var s = Array(1024).join('a') + '😀';
qs.stringify({ x: s }); // ended with %F0%9F%90%80%F2%90%80%80

After this change, the output correctly ends with %F0%9F%98%80.

Tests

• npx tape "test/**/*.js"
• npx tape test/utils.js
• npx eclint check lib/utils.js test/utils.js
• npx eslint lib/utils.js test/utils.js (no errors; existing no-continue warnings remain)
• verified dist/qs.js with the boundary reproduction