Enable testing wasm32-unknown-emscripten on travis

[malbarbo]

A disabled image was added in #42631. We should enable it when we get some capacity on travis.

[lilianmoraru]

@alexcrichton Can we re-enable wasm32 in Travis now?

[alexcrichton]

Unfortunately wasm32 is a "big enough" builder that it requires its own builder, which is much more of a significant investment than just having a standard library.

Did this regress though? I'd imagine that for all practical purposes having CI for asmjs is basically equivalent to CI for wasm

[lilianmoraru]

@alexcrichton In my case, the asserts were failing only on wasm32: #42518

[alexcrichton]

Was that ever working though? In the sense that would we have caught a regression if we were running CI?

[lilianmoraru]

@alexcrichton I was just pointing that asmjs and wasm32 are not equivalent - I myself hit an issue that was specific to wasm32 only.
I can't answer if this would of caught a regression, though.

[alexcrichton]

Oh yeah sure definitely!

I'm sure we'd catch bugs over time, but I'd personally sort of prefer to wait until we can make wasm a "true" tier 1 target by using the LLVM backend instead of going through emscripten, that may just take awhile :(

[alexcrichton]

This is done now for wasm32-unknown-unknown which I think is exercising LLVM's backend (the goal here), so I'm going to close.