Handle Unicode character casing and require Node.js 10

[sverweij]

what & why

Adds handling of non-ASCII character casing.

After merging this PR a string like розовый_пушистый-единороги will be converted to розовыйПушистыйЕдинороги (or to РозовыйПушистыйЕдинороги when pascal case is on) and Licorne époustouflante to licorneÉpoustouflante (/ LicorneÉpoustouflante).

Fixes #60

choices

Explanation of choices that might seem non-obvious:

• Replaces the \w with [\p{Alpha}\p{N}_] which is as close to the original intent of \w as I could muster (which is [A-Za-z0-9_] according to mdn)
• Uses unicode property escapes in most of the regular expressions as suggested in Correctly handle Unicode characters #60. This implies use of node >=10. I've updated the package.json and travis config to reflect this (+ added node stable to the build matrix).
• Keeps most of the existing logic in tact (if it ain't broken ...) except for this pattern in the preserveCamelCase function:

isLastCharLower && /[a-zA-Z]/.test(character) && character.toUpperCase() === character

isLastCharLower && /[\p{Lu}]/u.test(character)

Because the regular expression already implies it's an upper case character the character.toUpperCase() === character would always yield true anyway (but holler if I missed something!)

[sindresorhus]

This looks great 👌

Can you add an example to the readme and index.d.ts where it camelcases Unicode? Maybe just use the example from one of the tests.

Can you also mention at the top of the readme that it correctly handles Unicode?

[sverweij]

Will do - expect an update tonight (21:00 EST - so in ~9h time)