grapheme_* functions cannot catch differents between CJK Compatibility Ideographs and Unified Ideographs

Description

follow code

var_dump(grapheme_levenshtein("神", "神")); // 神 is U+FA19, but normalize to U+795E(神)

Actual result

int(0)

But expected result

int(1)

Because CJK Compatibility Ideographs keeps backword compatible, However, Normalized Compatible Ideograph(Unified Ideographs) users feels differents.
So I think create catch the CJK Compatible Ideographs one option in grapheme_* functions.

This behavior is ICU depends.

CJK Compatible Ideographs

U+F900-U+FA0B Pronuncation variants from KS X 1001:1998 (Korea)
U+FA0C-U+FA0D Duplicate characters from Big 5 (Taiwan)
U+FA0E-U+FA2D The IBM 32 compatibility ideographs (Japan)
U+FA2E-U+FA2F Korean compatibility ideographs (Korea)
U+FA30-U+FA6A JIS X 0213 compatibility ideographs (Japan)
U+FA30-U+UA6D ARIB compatibility ideographs (Japan)
U+FA70-U+FAD9 DPRK compatibility ideographs (North Korea)

How to catch different

1. Call ublock_getCode
2. Return value of ublock_getCode is judge to CJK Compatible Ideograph: UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS || UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT.