[YSQL] implement predicate pushdown to optimize basic range queries

[kmuthukk]

Sample table:

     CREATE TABLE IF NOT EXISTS user_actions(
           id     text,
           msg_id integer,
           msg    text,
           PRIMARY KEY(id, msg_id)
        )

Insert about 1000 msg_ids for each user. Confirm that user 'u-0-10' has 1000 rows.

postgres=# select count(*) from user_actions where id='u-0-10';
 count
-------
  1000
(1 row)

Time: 8.730 ms

Looking up a specific msg with equality predicate is pretty fast:

postgres=# select * from user_actions where id='u-0-10' and msg_id=10;
   id   | msg_id |     msg
--------+--------+-------------
 u-0-10 |     10 | msg--10--10
(1 row)

Time: 0.883 ms

Looking up a specific msg or few msgs with a range predicate seems to go scan all the 1000 rows of user 'u-0-10' (since time seems similar to the count(*) query above) instead of pushing the range predicate down to the DocDB layer, and hence much slower than expected:

postgres=# select * from user_actions where id='u-0-10' and msg_id>=10 and msg_id<11;
   id   | msg_id |     msg
--------+--------+-------------
 u-0-10 |     10 | msg--10--10
(1 row)

Time: 8.116 ms