[YSQL] IN query should use index

[frozenspider]

Given the schema:

CREATE TABLE elements (id int PRIMARY KEY, val int, grp int);
CREATE INDEX elements_idx ON elements (grp) INCLUDE (val);

EXPLAIN shows that only the first of the following cases involves index:

postgres=# EXPLAIN SELECT val FROM elements WHERE grp = 3;
                                    QUERY PLAN
----------------------------------------------------------------------------------
 Index Only Scan using elements_idx on elements  (cost=0.00..4.11 rows=1 width=4)
   Index Cond: (grp = 3)

postgres=# EXPLAIN SELECT val FROM elements WHERE grp IN (1,2,3);
                           QUERY PLAN
-----------------------------------------------------------------
 Foreign Scan on elements  (cost=0.00..103.75 rows=1000 width=4)
   Filter: (grp = ANY ('{1,2,3}'::integer[]))

postgres=# EXPLAIN SELECT val FROM elements WHERE grp > 1 AND grp < 2;
                           QUERY PLAN
-----------------------------------------------------------------
 Foreign Scan on elements  (cost=0.00..105.00 rows=1000 width=4)
   Filter: ((grp > 1) AND (grp < 2))

(BETWEEN, etc. are also worth checking)

For reference, in Postgres they all have plans like this:

fs=# EXPLAIN SELECT val FROM elements WHERE grp IN (1,2,3);
                                 QUERY PLAN
----------------------------------------------------------------------------
 Bitmap Heap Scan on elements  (cost=4.69..15.12 rows=31 width=4)
   Recheck Cond: (grp = ANY ('{1,2,3}'::integer[]))
   ->  Bitmap Index Scan on elements_idx  (cost=0.00..4.68 rows=31 width=0)
         Index Cond: (grp = ANY ('{1,2,3}'::integer[]))

[frozenspider]

Related issue: #995

[ndeodhar]

Duplicate of #1223 and #995

[bmatican]

Talked to @m-iancu earlier, maybe let's hold on this for a while @JDNdeveloper, until we realize if there's a bottleneck on being able to have more folks working on these types of optimizations.