faster printing of floats by divmod10() and others

Development Libraries
5

replaced the tests for NAN and INF with a single test if the binary exponent is 0xFF, as this indicates both

patch for print.cpp

// check IEEE754 bit layout to understand code below.
// note the exponent is a power of 2
byte getExponentByte(float number)
{
    uint8_t e = (*(((byte*) &number)+3) & 0x7F) << 1;
    if (*(((byte*) &number)+2) & 0x80) e++;
    return e;
}

size_t Print::printFloat(double number, uint8_t digits, uint8_t notation) 
{ 
    size_t n = 0;
    int exponent = 0;

    byte e = getExponentByte(number);
    // Recognize NAN and INF in single test 
    if (e == 0xFF) 
    {
        if (isnan(number))
        {
            n += print("nan");
            return n;
        }
        // if not NAN => INF
        if (number < 0.0) n += write('-');
        n += print("inf");
        return n;
    }

    // handle negative values
    if (number < 0.0)
    {
        n += write('-');
        number = -number;
    }

....

the time dropped from 720 to 704 gain ~2%,

I need to check the timer code in the test program, as the step size becomes relative too big.