Implement optional link-time cross stage optimization #1088

Job	Run time
Linux Backcompat	14m 54s
linux-asan (gcc, g++, Debug, -fsanitize=address)	3m 36s
linux-asan (gcc, g++, Debug, -fsanitize=thread)	12m 39s
emscripten	1m 54s
linux-asan (gcc, g++, Debug, -fsanitize=undefined)	8m 31s
android (ON)	1m 21s
iOS	20m 9s
android (OFF)	1m 6s
linux (clang, clang++, Debug)	3m 19s
macos (macos-14, clang, clang++, Debug)	4m 31s
linux (clang, clang++, Release)	3m 0s
macos-shared (macos-14, clang, clang++, Release)	4m 16s
windows (windows-2019, windows, Debug)	5m 40s
macos (macos-14, clang, clang++, Release)	4m 4s
linux (gcc, g++, Debug)	6m 33s
windows (windows-2019, windows, Release)	5m 55s
macos (macos-14, gcc, g++, Debug)	4m 5s
linux (gcc, g++, Release)	2m 48s
windows-shared (windows-2019, windows, Debug)	3m 4s
macos (macos-14, gcc, g++, Release)	4m 11s
windows-shared (windows-2019, windows, Release)	5m 57s
macos (macos-13, clang, clang++, Debug)	13m 32s
linux-shared (clang, clang++, Release)	3m 15s
linux-shared (gcc, g++, Release)	1m 44s
macos (macos-13, clang, clang++, Release)	8m 40s
macos (macos-13, gcc, g++, Debug)	12m 43s
macos (macos-13, gcc, g++, Release)	13m 25s
	2h 54m 52s

Provide feedback