Add tracing to rustc_codegen_spirv

[LegNeato]

Fixes #192

[LegNeato]

Open questions:

1. Do we want to do it in __rustc_codegen_backend() so we can trace very early?
2. Do we want to use our own env var? (I mirrored it off rustc and rustdoc, but I can see a case for reusing RUSTC_LOG)
3. Do we want a tracing feature that completely compiles tracing-related code out if we want? There is some non-trivial work in Replace printing codegen args with tracing #196 to generate info for the logs. It looks like tracing has a couple of knobs to query, so I gated code behind that but it will still be in the binary. I would lean towards compiling it in always without a feature, as that is what rustc does
4. Do we want a test? We can probably put an "I am alive" type log item and then run a passing compiletest asserting on the stderr output.

[LegNeato]

tokio-rs/tracing#3183 for nu-ansi-term version fix.

[LegNeato]

I'm not sure I agree with the lint...I am mirroring what rustc and rustdoc do themselves. Thoughts? Should I change it to only use codegen args? For now I have updated the lint to allow it.

[LegNeato]

tokio-rs/tracing#3184 for regex stuff.

[LegNeato]

I am not going to document this until a future PR where there is actually logging in place.

[LegNeato]

@Firestar99 any thoughts on the open questions in #193 (comment) ?

[Firestar99]

To most of these: I don't really know, as I haven't used the tracing feature, nor done more than 2 weeks of hacking in the actual compiler. Most of my contributions have been intrinsics, which don't need deep end compiler changes, or just rebasing stuff.

3. I don't think we really need a feature for the compiler

[LegNeato]

Ok, I am going to land these then and we can change later.