TokenStream: Pretty-printing for string and character literals is hard to read

[lambda-fairy]

When I try to println!() a TokenStream that contains a string literal, I get this back:

"\u{3c}\u{70}\u{3e}\u{3c}\u{62}\u{3e}\u{70}\u{69}\u{63}\u{6b}\u{6c}\u{65}\u{3c}\u{2f}\u{62}\u{3e}\u{62}\u{61}\u{72}\u{72}\u{65}\u{6c}\u{3c}\u{69}\u{3e}\u{6b}\u{75}\u{6d}\u{71}\u{75}\u{61}\u{74}\u{3c}\u{2f}\u{69}\u{3e}\u{3c}\u{2f}\u{70}\u{3e}"

That's much harder to read than the unescaped equivalent:

"<p><b>pickle</b>barrel<i>kumquat</i></p>"

I think we should at least render printable ASCII characters (except backslashes and quotes) without escaping, to make it easier for macro authors to debug their generated code.