py: Implement mp_format_float for doubles and use where appropriate

This allows using (almost) the same code for printing floats everywhere,
removes the dependency on sprintf and uses just snprintf and
applies an msvc-specific fix for snprintf in a single place so
nan/inf are now printed correctly.
diff --git a/py/objcomplex.c b/py/objcomplex.c
index 8a424f7..b790b05 100644
--- a/py/objcomplex.c
+++ b/py/objcomplex.c
@@ -37,10 +37,7 @@
 #if MICROPY_PY_BUILTINS_COMPLEX
 
 #include <math.h>
-
-#if MICROPY_FLOAT_IMPL == MICROPY_FLOAT_IMPL_FLOAT
 #include "py/formatfloat.h"
-#endif
 
 typedef struct _mp_obj_complex_t {
     mp_obj_base_t base;
@@ -53,33 +50,23 @@
     mp_obj_complex_t *o = o_in;
 #if MICROPY_FLOAT_IMPL == MICROPY_FLOAT_IMPL_FLOAT
     char buf[16];
-    if (o->real == 0) {
-        mp_format_float(o->imag, buf, sizeof(buf), 'g', 7, '\0');
-        mp_printf(print, "%sj", buf);
-    } else {
-        mp_format_float(o->real, buf, sizeof(buf), 'g', 7, '\0');
-        mp_printf(print, "(%s", buf);
-        if (o->imag >= 0 || isnan(o->imag)) {
-            mp_print_str(print, "+");
-        }
-        mp_format_float(o->imag, buf, sizeof(buf), 'g', 7, '\0');
-        mp_printf(print, "%sj)", buf);
-    }
+    const int precision = 7;
 #else
     char buf[32];
+    const int precision = 16;
+#endif
     if (o->real == 0) {
-        sprintf(buf, "%.16g", (double)o->imag);
+        mp_format_float(o->imag, buf, sizeof(buf), 'g', precision, '\0');
         mp_printf(print, "%sj", buf);
     } else {
-        sprintf(buf, "%.16g", (double)o->real);
+        mp_format_float(o->real, buf, sizeof(buf), 'g', precision, '\0');
         mp_printf(print, "(%s", buf);
         if (o->imag >= 0 || isnan(o->imag)) {
             mp_print_str(print, "+");
         }
-        sprintf(buf, "%.16g", (double)o->imag);
+        mp_format_float(o->imag, buf, sizeof(buf), 'g', precision, '\0');
         mp_printf(print, "%sj)", buf);
     }
-#endif
 }
 
 STATIC mp_obj_t complex_make_new(mp_obj_t type_in, mp_uint_t n_args, mp_uint_t n_kw, const mp_obj_t *args) {