Implements #3229 - Changed multiline string literals

[xixixao]

Implements #3229. Please scrutinize. trimAndEscapeLines might not be the best method name under the sun. All feedback welcome.

[xixixao]

Also implements #610 (duplicates).

[satyr]

Keep [\s\S]. [^] doesn't work in JScript.

[jashkenas]

Fantastic. This works equally well for strings with interpolations in them?

Perhaps formatString instead of trimAndEscapeLines?

[xixixao]

@jashkenas See the last test case, any additional formats to test for?

[xixixao]

I added some maybe too esoteric cases (let me know if that's overdoing it).

[lydell]

Typo: "multine" instead of "multiline"

[lydell]

Tests I feel are missing:

# Handle escaped backslashes correctly.
eq 'escaped backslash at EOL\\
    next line', 'escaped backslash at EOL\\ next line'
eq 'triple backslash\\\
    next line', 'triple backslash\\next line'

# Another way of preserving whitespace around line endings.
eq 'first line
    \   backslash at BOL', 'first line \   backslash at BOL'
eq 'first line\
    \   backslash at BOL', 'first line\   backslash at BOL'

# Edge case.
eq 'lone

        \

      backslash', 'lone backslash'

Or something like that.

[xixixao]

@lydell The "Another way of..." cases don't make sense to me, did you mean not to put the backslash in the formatted strings?

[xixixao]

@lydell If not, we have that behavior (all your test cases are now included).

[lydell]

I did mean to put the backslash in the formatted strings. Even though we don't do anything special to the backslashes in that case, the tests are still important: We must make sure that way of preserving whitespace around line endings really works. Tests also work a bit as documentation.

Looks like I missed one test case: Even moar backslashes. Infinitely many should work. I tried this on master:

'\\\\\\
  a'

Expected:

'\\\\\\ a';

Actual:

'\\\\\a';

[xixixao]

Ah.