Replace PointerMap with polyfill for regular Map

[stuartpb]

Ideally, this polyfill would be left out of this project altogether (to be handled by a dedicated polyfill like CoreJS), but that can be done later.

[stuartpb]

Just signed the CLA, not sure why the status hasn't updated.

[scottgonzalez]

I'm not sure what this actually gets us. We're still not using the standard Map API even with this update.

[scottgonzalez]

In fact, won't this just throw errors in any browser that has Map?

[stuartpb]

This isn't using the standard Map API? What's different?

[scottgonzalez]

Map#size is not a function.

[stuartpb]

Right - so the polyfill should implement a getter for size instead of a function. On that now.

[stuartpb]

size uses a getter now.

[stuartpb]

Fixed tests (I think, I'm not familiar with intern).

[stuartpb]

Derp, changing how existing Map() is looked up now.

[stuartpb]

I'm going to go ahead and remove the part of the set test that tests internal implementation details of the polyfill.

[stuartpb]

Okay, the only errors in the build now are the ones that are present in the current master.

[stuartpb]

So, since this is only needed to map pointers with (low) integer IDs for keys, why not just use a sparse Array to polyfill the Map?

[scottgonzalez]

We went with the sparse array implementation from gh-190.