Now the line for each statement will show the diff resulting from the
combination of `before_statement_effect` and `statement_effect`. It's
still possible to observe each in isolation via
`borrowck_graphviz_format = "two_phase"`.

including their unnormalized forms.
Add more documentation for them.

`InferCtxt` contains six structures within `RefCell`s. Every time we
create and dispose of (commit or rollback) a snapshot we have to
`borrow_mut` each one of them.

This commit moves the six structures under a single `RefCell`, which
gives significant speed-ups by reducing the number of `borrow_mut`
calls. To avoid runtime errors I had to reduce the lifetimes of dynamic
borrows in a couple of places.

This has negligible perf impact, but it does improve the code a bit.

* Only query the specialization graph of any trait once instead of once per
  impl
* Loop over impls only once, precomputing impl DefId and TraitRef

This was showing up in profiles. Also deduplicates the code to get just
the impl header span.

This was previously a future-compat warning that has since been turned into
hard error, but without actually removing all the code.

Avoids a call to `traits::overlapping_impls`, which is expensive.

…rCtxt, r=varkor

Reduce the number of `RefCell`s in `InferCtxt`.

`InferCtxt` contains six structures within `RefCell`s. Every time we
create and dispose of (commit or rollback) a snapshot we have to
`borrow_mut` each one of them.

This commit moves the six structures under a single `RefCell`, which
gives significant speed-ups by reducing the number of `borrow_mut`
calls. To avoid runtime errors I had to reduce the lifetimes of dynamic
borrows in a couple of places.

r? @varkor

…arkor

Improve performance of coherence checks

The biggest perf improvement in here is expected to come from the removal of the remaining rust-lang#43355 warning code since that effectively runs the expensive parts of coherence *twice*, which can even be visualized by obtaining a flamegraph:

![image](https://user-images.githubusercontent.com/5047365/74091959-e1f41200-4a8b-11ea-969d-2849d3f86c63.png)

…olnay

Make `num::NonZeroX::new` an unstable `const fn`

cc rust-lang#53718

These require `#[feature(const_if_match)]`, meaning they must remain unstable for the time being.

… r=Centril

Correctly parse `mut a @ b`

r? @Centril

Closes rust-lang#67861
Closes rust-lang#67926

…phviz, r=wesleywiser

Small graphviz improvements for the new dataflow framework

Split out from rust-lang#68241. This prints the correct effect diff for each line (the before-effect and the after-effect) and makes marginal improvements to the graphviz output for the new dataflow framework including using monospaced font and better line breaking. This will only be useful when rust-lang#68241 is merged and the graphviz output changes from this:

![image](https://user-images.githubusercontent.com/29463364/74107776-5e3c3300-4b28-11ea-9d33-4862228b5e87.png)

to this:

![image](https://user-images.githubusercontent.com/29463364/74107751-20d7a580-4b28-11ea-99ca-24f749386601.png)

r? @wesleywiser

parser: Keep current and previous tokens precisely

...including their unnormalized forms.
Add more documentation for them.

Hopefully, this will help to eliminate footguns like rust-lang#68728 (comment).

I'll try to address the FIXMEs in separate PRs during the next week.

r? @Centril