bugfix: floating point precision; added strutils.formatFloat

tests/accept/run/tfloat3.nim

@@ -0,0 +1,18 @@
+
+import math, strutils
+
+{.emit: """
+void printFloats(void) {
+    double y = 1.234567890123456789;
+    
+    printf("C double: %.10f, %.10f ", exp(y), cos(y));
+}
+""".}
+
+proc c_printf(frmt: CString) {.importc: "printf", header: "<stdio.h>", varargs.}
+proc printFloats {.importc, nodecl.}
+
+var x: float = 1.234567890123456789
+c_printf("Nimrod    %.10f, %.10f ", exp(x), cos(x))
+printFloats()
+