Use an integer cast to print floating point numbers more precisely

Improve printing six decimal places, remove trailing 0's, fix the carry when
numbers like 0.9999999 round up and print leading zeros in the decimal

The first attempt at printing floats had precision issues where the last few
digits would often be wrong. This next approach may yield a better algorithm
for numbers less than 4.29 billion, those that fit in 32 bits.

src/unity.c

@@ -284,6 +284,33 @@ void UnityPrintFloat(_UD number)
         _U_UINT exponent = 0;
         int scifmt;
         _U_UINT i;
+        _UU32 integer_part;
+        _UD fraction_part;
+        // if (number <= 0xFFFFFFFF) /* Fits in an integer */
+        {
+            integer_part = (_UU32)number;
+            fraction_part = number - integer_part;
+        }
+
+        _U_UINT fraction_bits = (_U_UINT)(fraction_part * 1000000.0f + 0.5f);
+        if (fraction_bits == 1000000)
+        {
+            fraction_bits = 0;
+            integer_part += 1;
+        }
+        _U_UINT divisor_int = 100000;
+
+        UnityPrintNumberUnsigned(integer_part);
+        UNITY_OUTPUT_CHAR('.');
+        /* now mod and print, then divide divisor */
+        do
+        {
+            UNITY_OUTPUT_CHAR((char)('0' + (fraction_bits / divisor_int)));
+            fraction_bits %= divisor_int;
+            if (fraction_bits == 0) break; // Truncate trailing 0's
+            divisor_int /= 10;
+        } while (divisor_int > 0);
+        UNITY_OUTPUT_CHAR(' ');
 
         while (number / divisor >= 10.0f)
         {