This has to re-introduce the `sink` pattern, because doing this purely
with iterators is awkward :( Maaaybe the event vector was a false start?
But, anyway, I like the current factoring more -- it sort-of obvious
that we do want to keep ws-attachment business in the parser, and that
we also don't want that to depend on the particular tree structure. I
think `shortcuts` module achieves that.
The general theme of this is to make parser a better independent
library.
The specific thing we do here is replacing callback based TreeSink with
a data structure. That is, rather than calling user-provided tree
construction methods, the parser now spits out a very bare-bones tree,
effectively a log of a DFS traversal.
This makes the parser usable without any *specifc* tree sink, and allows
us to, eg, move tests into this crate.
Now, it's also true that this is a distinction without a difference, as
the old and the new interface are equivalent in expressiveness. Still,
this new thing seems somewhat simpler. But yeah, I admit I don't have a
suuper strong motivation here, just a hunch that this is better.
The current implementation will throw a parser error for tuple structs
that contain a pub tuple field. For example,
```rust
struct Foo(pub (u32, u32));
```
is valid Rust, but rust-analyzer will throw a parser error. This is
because the parens after `pub` is treated as a visibility context.
Allowing a tuple type to follow `pub` in the special case when we are
defining fields in a tuple struct can fix the issue.