Retire outdated stage0 std remarks

[jieyouxu]

After the stage0 rework, our bootstrap chain is moreso the flat

precompiled                     in-tree
(stage0 rustc => stage0 std) => (stage1 rustc => stage1 std)

(where => here means "builds" or "is used to build")

as opposed to the chain previously

precompiled     in-tree (`cfg(bootstrap)`)             in-tree (`cfg(not(bootstrap))`)
stage0 rustc => stage0 std                          => (stage1 rustc => stage1 std)

so these library dances should no longer be needed, as we always build in-tree library with in-tree compiler, and the in-tree compiler will know about the new target.

[rustbot]

Thanks for the PR. If you have write access, feel free to merge this PR if it does not need reviews. You can request a review using r? rustc-dev-guide or r? <username>.

[jieyouxu]

cc @fee1-dead

[fee1-dead]

I will want to investigate whether the infra around STAGE0_MISSING_TARGETS is actually needed...

[jieyouxu]

I will want to investigate whether the infra around STAGE0_MISSING_TARGETS is actually needed...

For bootstrap it's a sanity check, since in parts of bootstrap we unfortunately have to make assumptions based on the existing target set (so other parts of the build system / test infra don't have to).

Ideally, the target specs would be moreso "pure definitions" that can be shared between bootstrap/test infra/compiler/library, but you know, it's a lot of work (design/impl wise). And ideally, you'd have "structured target spec query APIs" like being able to know which plausible target archs there are, etc. etc. (Custom targets throw a spanner in the works there.)