Fix rustdoc ICE for impl trait in argument position

[GuillaumeGomez]

Fixes #50358.

Weirdest fix ever...

r? @QuietMisdreavus

[rust-highfive]

The job x86_64-gnu-llvm-3.9 of your PR failed on Travis (raw log). Through arcane magic we have determined that the following fragments from the build log may contain information about the problem.

Click to expand the log.

[00:05:12] travis_fold:start:tidy
travis_time:start:tidy
tidy check
[00:05:12] tidy error: /checkout/src/test/rustdoc/ice-impl-trait-arg-position.rs:17: line longer than 100 chars
[00:05:12] tidy error: /checkout/src/test/rustdoc/ice-impl-trait-arg-position.rs:18: line longer than 100 chars
[00:05:14] some tidy checks failed
[00:05:14] 
[00:05:14] 
[00:05:14] command did not execute successfully: "/checkout/obj/build/x86_64-unknown-linux-gnu/stage0-tools-bin/tidy" "/checkout/src" "/checkout/obj/build/x86_64-unknown-linux-gnu/stage0/bin/cargo" "--no-vendor" "--quiet"
[00:05:14] 
[00:05:14] 
[00:05:14] failed to run: /checkout/obj/build/bootstrap/debug/bootstrap test src/tools/tidy
[00:05:14] Build completed unsuccessfully in 0:02:09
[00:05:14] Build completed unsuccessfully in 0:02:09
[00:05:14] Makefile:79: recipe for target 'tidy' failed
[00:05:14] make: *** [tidy] Error 1

The command "stamp sh -x -c "$RUN_SCRIPT"" exited with 2.
travis_time:start:00b1e897
$ date && (curl -fs --head https://google.com | grep ^Date: | sed 's/Date: //g' || true)

I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact @TimNN. (Feature Requests)

[KodrAus]

Now I don't feel so bad about making that exact same 'fix' in a local build so I could see my docs...

[GuillaumeGomez]

Then you did well. I really wonder why this assert was there...

[QuietMisdreavus]

I think the assumption behind that assert was that the closure passed to enter_impl_trait would clear out the cx.impl_trait_bounds map. If this assertion is being hit, i wonder if there's a pass somewhere that's not fully processing these bounds before they're getting dropped.

With this assertion removed, how does the documentation look for the included test? I'm curious whether it includes the impl Read deep in the bound there.

[GuillaumeGomez]

It does generate the documentation as expected. Which is why I wondered why such an assert was there...

[QuietMisdreavus]

It looks like the line (and function) in question was added in #49304 - @sinkuu, do you remember why you added this assert! in that PR? It's causing an ICE in the test that was added in this PR, and we're wondering what it's meant to catch.

[sinkuu]

The assertion error is legit. In this case, impl Read is not resolved (link to trait definition is not available). We need to correctly handle a synthesized type-parameter referring another.

[QuietMisdreavus]

Since @sinkuu opened #50419 with a proper fix, and incorporated the test from this PR, i'm closing this in favor of that one.