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/sql_explain.cc

@@ -1191,7 +1191,7 @@ const char *String_list::append_str(MEM_ROOT *mem_root, const char *str)
   if (!(cp = (char*)alloc_root(mem_root, len+1)))
     return NULL;
   memcpy(cp, str, len+1);
-  push_back(cp);
+  push_back(cp, mem_root);
   return cp;
 }