Our expected values for testcase "floating-point specifiers with 31-32 bit integer values" assume that %f specifiers are interpreted as %e for too many decimal digits; but - this assumption doesn't always hold. This it is an error to simply check that:

  PRINTING_CHECK("2.1474836470e+09", ==, sprintf_, buffer, "%.10f", 2147483647.0); // 2^31 - 1

and we should actually check for something different depending on MAX_INTEGRAL_DIGITS_FOR_DECIMAL.

Related to #111.