[clang] [llvm] [DTLTO][Clang] Add support for Integrated Distributed ThinLTO (PR #147265)

[Paul Kirth via llvm-commits]

ilovepi wrote:

> @Prabhuk I might need some help here. Looking at the output for the `Clang :: Driver/DTLTO/dtlto.c test` on your buildbot LLD is being invoked with the argument `"--thinlto-remote-compiler=/b/s/w/ir/x/w/llvm_build/bin/llvm"`. That is clearly not the location of a Clang compiler (`/b/s/w/ir/x/w/llvm_build/bin/clang`) ! This is produced by calling `ToolChain.getDriver().getClangProgramPath()` in the Clang driver (see: [5004c59#diff-c89ebf1fc950679f63c19739a35aa966a12674f7d61af6fb382ab1fa0ee21dd7R1328](https://github.com/llvm/llvm-project/commit/5004c59803fdeb3389d30439a6cc8b1ff874df0c#diff-c89ebf1fc950679f63c19739a35aa966a12674f7d61af6fb382ab1fa0ee21dd7R1328)). Is there something strange about the buildbot or perhaps the `fuchsia` target that might lead to this?

We use the multicall library. that is in fact our clang and lld (multicall is basically like a busybox config). 

The toolchain we're building and testing is just a normal linux/windows/mac toolchain. Its configured to compile Fuchsia, but we don't enable anything too special, just some standard cmake options.

https://github.com/llvm/llvm-project/pull/147265