ICU 75 removed the underscored UVS_* Unicode Variation Selector API
(UVS_combine_code / UVS_get_codepoint_length / UVS_get_codepoint_in_sequence)
in favor of the camelCase UVS* names that ICU exposed alongside them,
then ICU 78 removed the camelCase names too — variation-selector
handling is no longer part of the public ICU API. TL2025's upTeX
wrappers in kanji.h defined the camelCase names as macros expanding to
the underscored ones, so both build paths fail against modern ICU.

Replace the macros with no-op stubs returning 0. This unblocks the
texlive-core compile on ICU>=75 at the cost of upTeX losing
variation-selector handling. With USE=-cjk (default in this overlay)
the affected upTeX binaries (updvitype, uppltotf, uptftopl) are
built but not installed; with USE=cjk the binaries install but
variation-selector codepoints render as plain glyphs.

A proper fix would re-implement VS lookup against ICU's
StandardizedVariants table directly (uchar.h Variation_Selector
property), or vendor a small data table — both larger than this
overlay's scope. Filed upstream awaiting feedback.

--- a/texk/web2c/uptexdir/kanji.h
+++ b/texk/web2c/uptexdir/kanji.h
@@ -71,9 +71,9 @@
 #define inputline2(fp,buff,pos,size) input_line2(fp,buff,NULL,pos,size,NULL)
 #endif
 #define ptencconvfirstline(pos,limit,buff,size) ptenc_conv_first_line(pos,limit,buff,size)
-#define UVScombinecode(ucv,uvs) UVS_combine_code(ucv,uvs)
-#define UVSgetcodepointlength(ucv) UVS_get_codepoint_length(ucv)
-#define UVSgetcodepointinsequence(ucv,n) UVS_get_codepoint_in_sequence(ucv,n)
+static inline int UVScombinecode(int ucv, int uvs) { (void)ucv; (void)uvs; return 0; }
+static inline int UVSgetcodepointlength(int ucv) { (void)ucv; return 0; }
+static inline int UVSgetcodepointinsequence(int ucv, int n) { (void)ucv; (void)n; return 0; }
 #define ptencucsto8bitcode(enc,uch) ptenc_ucs_to_8bit_code(enc,uch)
 #define ptenc8bitcodetoucs(enc,ech) ptenc_8bit_code_to_ucs(enc,ech)