A cleanup for MDEV-19468 Hybrid type expressions return wrong format for FLOAT

Fixing problems revealed by buildbot:

- Fixing compilation failure on Windows
- Recoding correct engines/iuds/r/insert_decimal.result

mysql-test/suite/engines/iuds/r/insert_decimal.result

@@ -1151,7 +1151,7 @@ total_rows	min_value	max_value	sum	avg
 7	-100000.00000	100000.00000	-99.15000	-16.525000509
 SELECT count(*) as total_rows, min(c3) as min_value, max(c3) as max_value, sum(c3) as sum, avg(c3) as avg FROM t1;
 total_rows	min_value	max_value	sum	avg
-7	-0.10000000149011612	111111112	111211212.95000306	18535202.15833384
+7	-0.1	111111000	111211212.95000306	18535202.15833384
 SELECT count(*) as total_rows, min(c1) as min_value, max(c1) as max_value, sum(c1) as sum, avg(c1) as avg FROM t2;
 total_rows	min_value	max_value	sum	avg
 30	-10000000000	10000000000	31322222339	1044074077.9667
@@ -1160,7 +1160,7 @@ total_rows	min_value	max_value	sum	avg
 30	0	10000000000	43444444564	1448148152.1333
 SELECT count(*) as total_rows, min(c3) as min_value, max(c3) as max_value, sum(c3) as sum, avg(c3) as avg FROM t2;
 total_rows	min_value	max_value	sum	avg
-30	-3.4028234663852886e38	3.4028234663852886e38	1.0208470399155866e39	3.4028234663852886e37
+30	-3.40282e38	3.40282e38	1.0208470399155866e39	3.4028234663852886e37
 SELECT * FROM t1;
 c1	c2	c3	c4
 0.00000	-0.10000	-0.1	13

sql/item.cc

@@ -10168,7 +10168,7 @@ String* Item_cache_float::val_str(String *str)
   DBUG_ASSERT(fixed == 1);
   if (!has_value())
     return NULL;
-  Float((float) value).to_string(str, decimals);
+  Float(value).to_string(str, decimals);
   return str;
 }
 

sql/sql_type_real.h

@@ -23,7 +23,18 @@ class Float
 public:
   Float(float nr)
    :m_value(nr)
-  { }
+  {
+    DBUG_ASSERT(!std::isnan(nr));
+    DBUG_ASSERT(!std::isinf(nr));
+  }
+  Float(double nr)
+   :m_value((float) nr)
+  {
+    DBUG_ASSERT(!std::isnan(nr));
+    DBUG_ASSERT(!std::isinf(nr));
+    DBUG_ASSERT(nr <= FLT_MAX);
+    DBUG_ASSERT(nr >= -FLT_MAX);
+  }
   Float(const uchar *ptr)
   {
     float4get(m_value, ptr);