SQL remainder operation [IMDG-147]

[devozerov]

This PR introduces the remainder (mod), operator:

1. The operator coerces both operands to the type with the higher precedence without extending the bit width (unlike division) because it seems that there could be no overflow.
2. The return type is the same as the operand type. Alternatively, we may have reduced it to the second operand (e.g. BIGINT % SMALLINT => SMALLINT), but this would complicate the implementation of the operator because we would have many more combinations. We may leave it as a future optimization, but meanwhile, I prefer to keep things simple.
3. Inexact types (float, double) are not supported, like in major databases (e.g., Postgres, SQL Server).

Closes #17334

[petrpleshachkov]

Looks good to me!