[Release] hama v1.3.12: Character-span matching by default

Version 1.3.12 makes character-span pronunciation matching the default in both runtimes while keeping Python hama and TypeScript hama-js aligned on the same release.

• Default matching now ignores whitespace inside the candidate span and can match inside larger tokens while preserving original-input offsets and replacement behavior.
• Character-span matching now skips candidate windows that would exceed the G2P encoder input limit instead of silently truncating them.
• Kept explicit token-span behavior available for callers that still want token-boundary matching.
• Added Python and TypeScript test coverage for default matching of 성민 inside 성민님이, whitespace-insensitive 성 민 matching, explicit token-mode boundaries, and overlong-input guardrails.
• Aligned Python hama and TypeScript hama-js on version 1.3.12.