[Normative] Capture the HTML entity behaviors (#136)

## Summary

Let's be faithful to the de-facto and document the HTML entity behaviors to the spec. Note that this is not about whether we should "drop this semantics or not", but about documenting the current behaviors that everyone has been living with for years.

### The Proposed Normative Change

I'm not aware of any practices specifying such transpiler/transform semantics in ECMA-262 so this is a really interesting attempt 🙂 So I ended up extending `Static Semantics: SV` which is the smartest way I can find to hack the semantics into the ECMA-262 spec. I think this should work and should be accurate enough. I'm curious on how implementors think about it though.

<del>I also intentionally left the set of supported HTML entities implementation-defined to allow either HTML4 or HTML5 set. This may be seen as a breaking change in some regard and **this is open to discuss here**. </del> We've reached consensus that only HTML4 entities are allowed.

This commit also close #133 by using `::` for characters which are supposed to be lexical grammars.

Close #126
Close #4

## Test Plan

open `index.html` and proof-read the spec ;)