If a process operation results in less pages then the current set has, it will likely corrupt the page_map since it overwrites page_map entries ahead of where it's processing. This fixes that problem by removing page_map entries that will be dropped. Then dropping orphaned pages and re-indexing retained pages.
Fixes this -fno-sanitize-recover=undefined fail,
/set/iter: hb-algs.hh:1016:60: runtime error: index 4294967295 out of bounds for type 'unsigned long long const[8]'
#0 0x4d1e09 in hb_vector_size_t<unsigned long long, 64u>::operator[](unsigned int) const /home/user/code/harfbuzz/src/./hb-algs.hh:1016:60
#1 0x4d8b5f in hb_set_t::page_t::previous(unsigned int*) const /home/user/code/harfbuzz/src/./hb-set.hh:139:53
#2 0x4d0ada in hb_set_t::previous(unsigned int*) const /home/user/code/harfbuzz/src/./hb-set.hh:602:36
#3 0x4cd76f in hb_set_previous /home/user/code/harfbuzz/src/hb-set.cc:494:15
#4 0x4ca8af in test_set_iter /home/user/code/harfbuzz/test/api/test-set.c:310:3
#5 0x7f3a4f3e0f49 (/usr/lib/x86_64-linux-gnu/libglib-2.0.so.0+0x72f49)
#6 0x7f3a4f3e0e7a (/usr/lib/x86_64-linux-gnu/libglib-2.0.so.0+0x72e7a)
#7 0x7f3a4f3e1121 in g_test_run_suite (/usr/lib/x86_64-linux-gnu/libglib-2.0.so.0+0x73121)
#8 0x7f3a4f3e1140 in g_test_run (/usr/lib/x86_64-linux-gnu/libglib-2.0.so.0+0x73140)
#9 0x4c8894 in hb_test_run /home/user/code/harfbuzz/test/api/./hb-test.h:88:10
#10 0x4c8894 in main /home/user/code/harfbuzz/test/api/test-set.c:408:10
#11 0x7f3a4e3d2b96 in __libc_start_main /build/glibc-OTsEL5/glibc-2.27/csu/../csu/libc-start.c:310
#12 0x41e7d9 in _start (/home/user/code/harfbuzz/test/api/test-set+0x41e7d9)
This reverts commit c4aa10ebc8.
Broke several compilers... Sigh. The version without constexpr
didn't fully optimize out the unreachable code on clang.
So, revert it is...
Was good idea, but with C++ types with constructor/destructor, was getting in
the way as compiler was destructing those items where it was not desired.
Since C++ does not allow zero-sized arrays, just remove it...