Wording RegExp DecimalEscape is unnecessarily complicated

[hashseed]

In [0] for DecimalEscape, its evaluation is described: it can be either the NUL character or an integer.
In [1] for AtomEscape, the usage of that value in an atom is described: if it's a character, it describes that character. If it's an integer, it describes a back reference.
In [2] for ClassEscape, the usage of that value in a character class is described: if it's a character, it describes that character. If it's an integer, a syntax error is thrown.

It seems to me that the production for \0 and is unnecessarily entwined. In unicode mode, the former exclusively describes the character NUL and the latter a back reference (if matching capturing group exists), since capturing groups are 1-indexed. In non-unicode mode, both could also fall back to be a legacy octal literal, for which a separate syntax pattern already exists.

If we factor out \0 as NullEscape, we could

• add it alongside DecimalEscape in the production of AtomEscape.
• remove DecimalEscape in the production of ClassEscape.
• remove step 2 in [1].
• remove step 2 in [2].

It seems to me that this would make things a lot easier to understand, while preserving the way things work.

[0] http://tc39.github.io/ecma262/#sec-decimalescape
[1] http://tc39.github.io/ecma262/#sec-atomescape
[2] http://tc39.github.io/ecma262/#sec-classescape