JavaScript read to the standard: number formats, and mistakes caught up front

A broad step up in how faithfully the engine reads JavaScript, so more of the web's code runs exactly the way its authors intended. The engine now understands number formats it used to reject outright, including binary and octal numbers and the digit separators that make long numbers readable, and it reads every kind of text, regular expression and number the way the language standard describes. Just as important, it now reports a malformed piece of code the moment a page is read, rather than partway through running it, which is where a correct browser is expected to raise such problems: an unfinished piece of text, a broken escape, an impossible regular expression, or a number written the wrong way are all flagged up front instead of failing halfway through and leaving a page half-built. The built-in objects the language provides were also brought in line with the standard, so the information they report about themselves and the way their contents are listed now match what real libraries quietly rely on. None of this changes code that was already correct: it makes the engine agree with other browsers across a large body of fine details that real-world code depends on. The JavaScript language aspect moves 86 to 88; the overall platform score stays at 70.