MDEV-7951 - sql_alloc() takes 0.25% in OLTP RO

sql_alloc() has additional costs compared to direct mem_root allocation:
- function call: it is defined in a separate translation unit and can't be
  inlined
- it needs to call pthread_getspecific() to get THD::mem_root

It is called dozens of times implicitely at least by:
- List<>::push_back()
- List<>::push_front()
- new (for Sql_alloc derived classes)
- sql_memdup()

Replaced lots of implicit sql_alloc() calls with direct mem_root allocation,
passing through THD pointer whenever it is needed.

Number of sql_alloc() calls reduced 345 -> 41 per OLTP RO transaction.
pthread_getspecific() overhead dropped 0.76 -> 0.59
sql_alloc() overhed dropped 0.25 -> 0.06

sql/opt_range.h

@@ -69,13 +69,13 @@ class QUICK_RANGE :public Sql_alloc {
   uint16 dummy;					/* Avoid warnings on 'flag' */
 #endif
   QUICK_RANGE();				/* Full range */
-  QUICK_RANGE(const uchar *min_key_arg, uint min_length_arg,
+  QUICK_RANGE(THD *thd, const uchar *min_key_arg, uint min_length_arg,
               key_part_map min_keypart_map_arg,
 	      const uchar *max_key_arg, uint max_length_arg,
               key_part_map max_keypart_map_arg,
 	      uint flag_arg)
-    : min_key((uchar*) sql_memdup(min_key_arg,min_length_arg+1)),
-      max_key((uchar*) sql_memdup(max_key_arg,max_length_arg+1)),
+    : min_key((uchar*) thd->memdup(min_key_arg, min_length_arg + 1)),
+      max_key((uchar*) thd->memdup(max_key_arg, max_length_arg + 1)),
       min_length((uint16) min_length_arg),
       max_length((uint16) max_length_arg),
       flag((uint16) flag_arg),