Add negative tests for legacy octal escape sequences in strings

[ajklein]

Annex B.1.2 defines legacy octal escape sequences in string literals. From what I can tell, test262 doesn't have coverage of these, either for their acceptance in sloppy mode code nor for them being disallowed in strict mode code (as Section 11.8.4 requires).

Thanks to the reporter of this v8 issue for bringing this to my attention.

[bakkot]

There's four things to be covered here: octal escapes (\01 etc) and non-octal escapes (\08 and \09), each in sloppy and strict code.

In sloppy code both are accepted but with different semantics (LegacyOctalEscapeSequence:: ZeroToThree OctalDigit [lookahead ∉ OctalDigit] and LegacyOctalEscapeSequence :: OctalDigit [lookahead ∉ OctalDigit] respectively). In strict code both are rejected (EscapeSequence :: 0 [lookahead ∉ DecimalDigit]).

I believe there's coverage for octal/sloppy and noctal/sloppy here, and coverage for octal/strict here (edit: and here).

The only missing case is noctals in strict code: that is, we don't test that '\08' is forbidden in strict code. (That's also, incidentally, the only case v8 gets wrong.)