1. Add missing lock / unlock in profiling code. Without
   synchronization, the profiling code could access freed memory.
2. Remove unnecessary assert.
3. Remove unncessary if check in taosHashRemoveWithData.

* Testing
sample mysql:
taos> create database db;
Query OK, 0 of 0 row(s) in database (0.002291s)

taos> use db;
Query OK, 0 of 0 row(s) in database (0.000197s)

taos> create table t (ts timestamp, a int);
Query OK, 0 of 0 row(s) in database (0.003851s)

taos> insert into t values ('2019-07-15 00:00:00', 1);
Query OK, 1 of 1 row(s) in database (0.001128s)

taos> insert into t values ('2019-07-15 01:00:00', 2);
Query OK, 1 of 1 row(s) in database (0.000249s)

taos> select * from t;
           ts            |      a      |
========================================
 2019-07-15 00:00:00.000 |           1 |
 2019-07-15 01:00:00.000 |           2 |
Query OK, 2 row(s) in set (0.000772s)

taos> drop database db;
Query OK, 0 of 0 row(s) in database (0.002308s)

1. Simplify implementation of taosHashCondTraverse
2. Complete function header comments for a few more functions

* Testing

Run through the sample mysql:
taos> create database db;
Query OK, 0 of 0 row(s) in database (0.002475s)

taos> use db;
Query OK, 0 of 0 row(s) in database (0.000252s)

taos> create table t (ts timestamp, a int);
Query OK, 0 of 0 row(s) in database (0.003588s)

taos> insert into t values ('2019-07-15 00:00:00', 1);
Query OK, 1 of 1 row(s) in database (0.001268s)

taos> insert into t values ('2019-07-15 01:00:00', 2);
Query OK, 1 of 1 row(s) in database (0.000265s)

taos> select * from t;
           ts            |      a      |
========================================
 2019-07-15 00:00:00.000 |           1 |
 2019-07-15 01:00:00.000 |           2 |
Query OK, 2 row(s) in set (0.000822s)

taos> drop database db;
Query OK, 0 of 0 row(s) in database (0.001711s)

A few cleanups:
1. Move implementation details (macro definition, SHashEntry and SHashObj
   definition) to hash.c. Only leave relevant APIs in hash.h.
2. Complete function header comments.
3. Correct variable naming from "newSize" to "newCapacity".

A few cleanups:
1. Rename hash node "count" to "refCount" for ease of reading
2. Trailing white space removal of hash.h

[td-5407]<enhance>: add a local buffer for super table vgroup id list to improve the query performance.