Fix substrReplace to be multibyte safe

[Nasim25]

Description

Fixes Str::substrReplace() to properly handle multibyte UTF-8 characters.

Problem

Current implementation uses substr_replace() which operates on bytes, causing corruption with multibyte characters:

Str::substrReplace('kenkä', 'ng', -3, 2);
// Current: "kenng¤" ❌
// Expected: "kenga" ✅

Solution

• Replaced substr_replace() with mb_substr() concatenation
• Properly normalizes negative offsets and lengths
• Maintains full backward compatibility

Testing

Str::substrReplace('kenkä', 'ng', -3, 2);  // 'kengä' ✅
Str::substrReplace('kenka', 'ng', -3, 2);  // 'kenga' ✅

Added tests for multibyte string handling, and tests pass.