Confirmed, the bot is happy again.
So was that a fresh regression or an old bug?

Found by the libFuzzer bot.
Probably a recent regression, please confirm.

==4356==ERROR: AddressSanitizer: heap-use-after-free on address 0x61d00001df68 at pc 0x0000006b9280 bp 0x7ffd94b57b30 sp 0x7ffd94b57b28
READ of size 8 at 0x61d00001df68 thread T0
    #0 0x6b927f in af_latin_hints_compute_segments src/autofit/aflatin.c:1537:67
    #1 0x6a33ec in af_latin_metrics_init_widths src/autofit/aflatin.c:183:17
    #2 0x6a33ec in af_latin_metrics_init src/autofit/aflatin.c:1009
    #3 0x6b138a in af_face_globals_get_metrics src/autofit/afglobal.c:482:17
    #4 0x6b138a in af_loader_load_glyph src/autofit/afloader.c:508
    #5 0x6b138a in af_autofitter_load_glyph src/autofit/afmodule.c:449
    #6 0x51e6a2 in FT_Load_Glyph src/base/ftobjs.c:733:19
    #7 0x4e49f0 in LLVMFuzzerTestOneInput

0x61d00001df68 is located 2280 bytes inside of 2288-byte region [0x61d00001d680,0x61d00001df70)
freed by thread T0 here:
    #0 0x4b6a8e in realloc /usr/local/google/home/kcc/llvm/projects/compiler-rt/lib/asan/asan_malloc_linux.cc:71:3
    #1 0x5238b7 in ft_mem_qrealloc src/base/ftutil.c:155:16
    #2 0x5238b7 in ft_mem_realloc src/base/ftutil.c:102

previously allocated by thread T0 here:
    #0 0x4b673b in __interceptor_malloc /usr/local/google/home/kcc/llvm/projects/compiler-rt/lib/asan/asan_malloc_linux.cc:52:3
    #1 0x52391d in ft_mem_qalloc src/base/ftutil.c:76:15
    #2 0x52391d in ft_mem_alloc src/base/ftutil.c:55
    #3 0x52391d in ft_mem_qrealloc src/base/ftutil.c:146
    #4 0x52391d in ft_mem_realloc src/base/ftutil.c:102