Result-ify `src/parser.rs`

[alexcrichton]

This commit converts all of src/parser.rs away from panics to using
Diagnostic instead. Along the way this adds a test case per changed panic!,
ensuring that we don't regress in these areas!

cc #601

[fitzgen]

How does one run the UI tests? Is it the same as all of our other tests? If not, we should also document that.

Also, are rustc error messages stable enough that we can rely on this stderr content remaining the same?

Overall, this is a huge improvement for the UX of using wasm-bindgen! Thanks, Alex!

[fitzgen]

How does one run the UI tests? Is it the same as all of our other tests? If not, we should also document that.

Also, are rustc error messages stable enough that we can rely on this stderr content remaining the same?

Overall, this is a huge improvement for the UX of using wasm-bindgen! Thanks, Alex!

[alexcrichton]

How does one run the UI tests?

We've got some handy documentation about that!

Is it the same as all of our other tests?

Indeed! A simple cargo test -p ui-tests should do the trick.

Also, are rustc error messages stable enough that we can rely on this stderr content remaining the same?

I believe that since all this output is our error messages it should hopefully be pretty stable. That being said I wouldn't be surprised if these tests break over time. I'd like to eventually add UI tests as well for failing conditions like "you specified an argument that didn't implement IntoWasmAbi" to make sure it looks reasonable, but that sort of error message is practically guaranteed to change over time.

I think we may eventually move this test suite into an "allow failures" category on Travis and only update it every now and then (such as when the test suite itself is modified), but for now we hopefully won't have too many issues!