Added type annotations for a few glue code functions

[RunDevelopment]

I added type annotations to a few commonly-used JS glue code functions. This makes it a lot easier to understand the glue, because I have TS to help me out.

Those type annotations also make it easier to verify the correctness of glue. And just to prove that: TS found 2 minor bugs in the JS glue code of our tests:

Fixes moved to #4192.

1. debugString (this is the JS glue code version of format!("{:?}", value)) had a bug where it checked for a regex mismatch incorrectly. This would lead to dereferencing null at runtime, which throws a TypeError.

2. _assertClass returned instance.ptr. Here's the full code of the function:

   function _assertClass(instance, klass) {
       if (!(instance instanceof klass)) {
           throw new Error(`expected instance of ${klass.name}`);
       }
       return instance.ptr;
   }

   Returning instance.ptr is weird for 2 reasons:

   • It's not used anywhere in the code. _assertClass is only used in one place and its return value is ignored.
   • There is no ptr field. The author probably meant __wbg_ptr to get the internal pointer of a JS object.

   So I just removed the return statement and instance.ptr.

---

Aside from that, I did one additional change: I inline lTextDecoder:

// before

const lTextDecoder = typeof TextDecoder === 'undefined' ? (0, module.require)('util').TextDecoder : TextDecoder;

let cachedTextDecoder = new lTextDecoder('utf-8', { ignoreBOM: true, fatal: true });

// after

/** @type {TextDecoder} */
let cachedTextDecoder = new (typeof TextDecoder === 'undefined' ? (0, module.require)('util').TextDecoder : TextDecoder)('utf-8', { ignoreBOM: true, fatal: true });

This just made typing easier, but the expression something to get used to...

[daxpedda]

I am in favor of this change, but I don't feel comfortable approving this without any further input, ideally approval from another maintainer.
Cc @Liamolucko.

In the meantime, I think it would be good to split off the bugfixes you made into a separate PR.
I would also appreciate if you could do some code archeology into instance.ptr and see where this originated and the context around it.

Didn't review it yet, but it needs a changelog entry.

[daxpedda]

This is amazing work by the way, thank you!

[RunDevelopment]

I think it would be good to split off the bugfixes you made into a separate PR.

Will do.

I would also appreciate if you could do some code archeology into instance.ptr and see where this originated and the context around it.

Seems like the pointer field changed names a few times. 240d3cd (7 years ago) changed it from __wasmPtr to ptr and dbea2a2 (1 year ago) changed it from ptr to __wbg_ptr. So it's not that the author of _assertClass used the wrong field name, the code of _assertClass just wasn't updated when the pointer field was changed to the current __wbg_ptr.

[daxpedda]

I also found 5d697c1, which was the point where the return value stopped being used anymore but didn't get removed.

[daxpedda]

@RunDevelopment I'm happy to go ahead and merge this after a rebase.

Would that solve the requirements for #4311 to run type checks on JS as well?