ascii=true and fullhex=true flags for escape_string

[stevengj]

This PR adds two new optional keyword flags ascii=true and fullhex=true to the escape_string function, both of which default to false (= current behavior).

If ascii=true is passed, then all non-ASCII characters are escaped. If fullhex=true is passed, then \u and \U escapes are printed with 4- and 8-digit hex values, respectively (instead of omitting leading zeros).

Motivation:

• I often find myself wanting to escape non-ASCII characters in order to see more easily what codepoints a string contains, both for debugging/pedagogy and for literal strings in code where character normalization is nonobvious. For example, if you want to be explicit about whether the string "äöü" is written in NFC normalization "\u00e4\u00f6\u00fc" or NFD normalization "a\u0308o\u0308u\u0308", and don't want to run the risk of an editor accidentally re-normalizing the string for you (as happens in some browsers), you have to write it escaped.
• printing the full 4/8-digit hex values is important if you want to print the string in a form that is compatible with C or other C-like languages.

[fingolfin]

Looks good to me