Add assertions to emscriptenGetAudioObject. NFC

[sbc100]

This API was added when audioworklets were first added, but it doesn't
look like it was intended to work with invalid inputs: 5402fc9.

At least none of the original usages in 5402fc9 checked for

[juj]

The above call to emAudioExpectNodeOrContext() seems restrictive.

It prevents users from testing if a given handle has been deleted? E.g. one would not be able to write code like this anymore:

if (emscriptenGetAudioObject(someHandle)) {
  console.log('someHandle exists, accessing it...');
} else {
  console.log('someHandle has been deleted, doing something else (maybe deinitializing some JS side structure at shutdown?)');
}

[sbc100]

Indeed, that is the idea. None of our use cases for the API seems to use it in this way, and none of the other APIs in the this file allow invalid handles to be passed. Unless there is specific use case for allowing invalid handles here I suggest we be consistent. (At least until someone asks for this ability).

[cwoffenden]

I'd be interested to know how others are using the API. I agree with putting the test with the handle request, but I only see my own use cases.

[sbc100]

I'm only aware of the uses within the test code, but maybe you could find more via a github global search or some such?

[cwoffenden]

but maybe you could find more via a github global search or some such?

Searching GH for emscriptenGetAudioObject( shows only 278 matches, and nearly all of the code will fail due to never checking for a non-null return (e.g. emscriptenGetAudioObject($0).connect). I did excitedly see some code that did check, and it was code I'd written for the audio tests.