StringLowering: Escape the JSON in the custom section

[kripken]

Also add an end-to-end test using node to verify we can parse the escaped content properly using TextDecoder+JSON.parse.

[kripken]

PR now contains code point logic from Emscripten, that is hopefully correct... on a large real-world testcase it seems to parse everything properly at least.

[kripken]

I ported the warning from the JS code to be more defensive here, and I added more comments in general. Did you have anything else in mind for defense/readability?

I added test coverage for all escaped characters.

[kripken]

Last commit adds a test for a weird utf8 char, and fixes our handling of the escape codes.

[tlively]

I plan to try to rewrite this to be safer and easier to understand, so LGTM as-is for the short term.