Refactor scopes tree decoding

[nicolo-ribaudo]

See tc39#196 (comment).

I tried to make it a bit easier to see what is going on by:

• giving different names to the different SDOs that take care of parsing different data about the scopes
• getting all the data and then creating the Original Scope Record, rather than creating an Original Scope Record and then mutating it
• centralizing the "increment the index in the state and get the result" in a single AO

We could do something similar for TopLevelItemList.

[szuend]

Since we need to set the result on DecodeOriginalScopeTrees now on _info_.[[Scopes]], do we also need to duplicate this for the production:

Scopes :
  OriginalScopeTreeList

Since the application of the chain rule only performs the SDO but does not set t on _info_?

[szuend]

It is important that we preserve the OriginalScopeTreeItem: [empty] case with a null.

The motivation is that if a source map has a list of N sources, then we expect N original scope trees. To signify that a specific source file doesn't have an original scope tree, we currently use the empty production and add null.

The PR currently removes the null, so afterwards we can't associate source records with their respective original scope tree.

I don't have a strong opinion if we want to do this differently. Either by having an "empty" tag, or by adding a "sourceFileIndex" field or item.

If we keep the "null" route we'd need to:

• Return here a List of either a Original Scope Record or ~null~
• Change the empty production to return a single element list: 1. Return « ~null~ »

[nicolo-ribaudo]

Should we also add null to the flattened scopes?

[szuend]

I'd say no. Reason being that the "definition index" in a generated range is basically the Nth "original scope start" item. I.e. an index into the flattened scopes without pushing the top-level null scopes.

[nicolo-ribaudo]

I have an extra closing tag somewhere, that's why the diff is weird.

[nicolo-ribaudo]

Fixed!

[nicolo-ribaudo]

I'll re-open on the other repo