Clean up the lto_cflags mess

We were defining lto_cflags in three places. It looks like on
aarch64 nil won (LTO disabled), while on other architectures
`-flto=thin` won.

I think on rawhide, this means that we're currently shipping
bitcode in static archives, because the required
`-ffat-lto-objects` option from redhat-rpm-config is missing.

Resolve this by leaving the option at its default value --
if enabling it on aarch64 causes issues, we can disable it just
there again.

Also reenable the LLVM_UNITTEST_LINK_FLAGS option, which should
reduce the amount of time LTO takes.

build.spec.inc

@@ -1,8 +1,3 @@
-# Disable LTO to speed up builds
-%if %{with snapshot_build}
-%global _lto_cflags %nil
-%endif
-
 # TODO(kkleine): In clang we had this %ifarch s390 s390x %{arm} aarch64 %ix86 ppc64le
 %ifarch s390 s390x %{arm} %ix86
 # Decrease debuginfo verbosity to reduce memory consumption during final library linking
@@ -17,11 +12,6 @@ export ASMFLAGS="%{build_cflags}"
 
 #region BEGIN: COPIED FROM CLANG
 
-# And disable LTO on AArch64 entirely.
-%ifarch aarch64
-%define _lto_cflags %{nil}
-%endif
-
 # TODO(kkleine): Make sure the paths actually work. I have the feeling this needs adjustment
 pwd
 %if 0%{?__isa_bits} == 64
@@ -203,9 +193,7 @@ cd llvm
 %ifarch x86_64
 	-DCMAKE_SHARED_LINKER_FLAGS="$LDFLAGS -Wl,-z,cet-report=error" \
 %endif
-%if %{without snapshot_build}
 	-DLLVM_UNITTEST_LINK_FLAGS="-Wl,-plugin-opt=O0"
-%endif
 
 # Build libLLVM.so first.  This ensures that when libLLVM.so is linking, there
 # are no other compile jobs running.  This will help reduce OOM errors on the

globals.spec.inc

@@ -23,8 +23,6 @@
 %global src_tarball_dir llvm-project-%{llvm_snapshot_git_revision}
 %endif
 
-%global _lto_cflags -flto=thin
-
 #region LLVM globals
 
 %if %{with compat_build}