Aggregate function discrepancies on empty or null datasets

[Ramkarthik]

The following aggregate functions have discrepancies when applied on empty datasets or only-null values.

Steps to reproduce (create empty table since limbo doesn't support where clause yet): create table empty_table (id INT PRIMARY KEY);

AVG

SQLite:

> select avg(id) from empty_table;

Limbo:

> select avg(id) from empty_table;
thread 'main' panicked at core/vdbe.rs:609:33:
internal error: entered unreachable code
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

COUNT

SQLite:

> select count(id) from empty_table;
0

Limbo:

> select count(id) from empty_table;
NULL

The ones below have differences, but I don't know if we want to change or not.

SUM

SQLite:

> select sum(id) from empty_table;

Limbo:

> select sum(id) from empty_table;
NULL

MIN

SQLite:

> select min(id) from empty_table;

Limbo:

> select min(id) from empty_table;
NULL

MAX

SQLite:

> select max(id) from empty_table;

Limbo:

> select max(id) from empty_table;
NULL

[benclmnt]

SUM, MIN, MAX looks like a display issue. I will take a look at AVG and COUNT