Inline WorkerThread pointer into a thread local

[zetanumbers]

I've got this idea while implementing #155649. It's not as clean as I've expected but it did get rid of many unsafe blocks. I can also remove these unsafe blocks without changing WORKER_THREAD_STATE's type by utilizing new safe WorkerThread functions.

[rustbot]

r? @chenyukang

rustbot has assigned @chenyukang.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler
• compiler expanded to 73 candidates
• Random selection from 19 candidates

[zetanumbers]

r? @nnethercote

[zetanumbers]

Doesn't seem to influence the performance overall, but noise level is a bit too big on my machine. Perhaps @Zoxc has better environment.

Benchmark	base~main~6	new~worker_thread_state~2
	Time	Time	%
🟣 hyper:check	0.1121s	0.1105s	💚 -1.45%
🟣 hyper:check:initial	0.1219s	0.1279s	💔 4.92%
🟣 hyper:check:unchanged	0.0744s	0.0771s	💔 3.73%
🟣 clap:check	0.3701s	0.3723s	0.58%
🟣 clap:check:initial	0.4354s	0.4341s	-0.31%
🟣 clap:check:unchanged	0.2227s	0.2237s	0.45%
🟣 syn:check	0.4147s	0.4190s	💔 1.04%
🟣 syn:check:initial	0.5133s	0.5073s	💚 -1.15%
🟣 syn:check:unchanged	0.3295s	0.3299s	0.09%
🟣 regex:check	0.2623s	0.2557s	💚 -2.53%
🟣 regex:check:initial	0.3098s	0.3063s	💚 -1.13%
🟣 regex:check:unchanged	0.1821s	0.1836s	0.79%
Total	3.3483s	3.3473s	-0.03%
Summary	1.0000s	1.0042s	0.42%

[zetanumbers]

Also I haven't yet updated comments. TODO

[Zoxc]

I was thinking a scoped thread local would make sense for this too. cc @cuviper in case that makes sense for upstream Rayon.

My guess is that the indirection doesn't have much performance overhead. A Rayon benchmark would be more suitable, but not sure it would pick that up.