invariant: GCB=hst for relevant values

[markusicu]

Hangul_Syllable_Type values correspond to Grapheme_Cluster_Break values where hst!=NA, except for some Kirat Rai vowel signs (new in Unicode 16). [:GCB=LV:] == [:hst=LV:] etc.

In ICU code (uprops.cpp):

/*
 * Map some of the Grapheme Cluster Break values to Hangul Syllable Types.
 * Hangul_Syllable_Type is fully redundant with a subset of Grapheme_Cluster_Break.
 *
 * Starting with Unicode 16, this is not quite true:
 * Some Kirat Rai vowels are given GCB=V for proper grapheme clustering, but
 * they are of course not related to Hangul syllables.
 */
static const UHangulSyllableType gcbToHst[]={
    U_HST_NOT_APPLICABLE,   /* U_GCB_OTHER */
    U_HST_NOT_APPLICABLE,   /* U_GCB_CONTROL */
    U_HST_NOT_APPLICABLE,   /* U_GCB_CR */
    U_HST_NOT_APPLICABLE,   /* U_GCB_EXTEND */
    U_HST_LEADING_JAMO,     /* U_GCB_L */
    U_HST_NOT_APPLICABLE,   /* U_GCB_LF */
    U_HST_LV_SYLLABLE,      /* U_GCB_LV */
    U_HST_LVT_SYLLABLE,     /* U_GCB_LVT */
    U_HST_TRAILING_JAMO,    /* U_GCB_T */
    U_HST_VOWEL_JAMO        /* U_GCB_V */
    /*
     * Omit GCB values beyond what we need for hst.
     * The code below checks for the array length.
     */
};

static int32_t getHangulSyllableType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
    // Ignore supplementary code points: They all have HST=NA.
    // This is a simple way to handle the GCB!=hst cases since Unicode 16 (Kirat Rai vowels).
    if(c>0xffff) {
        return U_HST_NOT_APPLICABLE;
    }
    /* see comments on gcbToHst[] above */
    int32_t gcb=(int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_GCB_MASK)>>UPROPS_GCB_SHIFT;
    if(gcb<UPRV_LENGTHOF(gcbToHst)) {
        return gcbToHst[gcb];
    } else {
        return U_HST_NOT_APPLICABLE;
    }
}