feat(v4): add z.script() for unicode script validation

[FucciUnavailable]

Will eventually close #5804

This PR adds z.script(name) which validates that every character in the string belong to the given unicode script, using the \p{Script=lang} property escape built into the JS engine

Expected behavior:

z.script("Arabic").parse("حفلة"); // passes
z.script("Latin").parse("hello"); // passes
z.script("Cyrillic").parse("привет"); // passes

z.string().script("Cyrillic") // the method form also available and passed the tests

z.script("NotAScript") // throws syntax error at Schema creation; The JS Engine rejects unknowin script identifiers.

Notes:

• Invalid script names fail fast at schema creation time (not parse time) => the JS engine rejects unknown \p{Script=...} identifiers immediately

• Strict by design: spaces, punctuation, and diacritics outside the script's own code points will fail (suitable for validating individual words/tokens, not full sentences) => Support can be extended with Script_Extensions instead of Script ^\p{Script_Extensions=Arabic}+$

• Script names are the official Unicode identifiers (Old_Persian, Linear_B..etc) => casing matters

• New scripts are supported automatically as Node.js/V8 updates their ICU Unicode tables => nothing to maintain in Zod

Motivation:

Making zod even more accessible worldwide! :D

Important:

Named aliases (z.arabic(), z.latin(), etc.) are intentionally left out each would be a one-liner in classic/schemas.ts, but picking which of the 150+ Unicode scripts deserve a named helper is an arbitrary call.
z.script("Arabic") is already readable, and the escape hatch is there if we want to ship named aliases later.

Reference Issue: #5804