Decimal overflow misbehavior in BinaryArithmeticFunctions.

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

use test;
CREATE TABLE t
(pk INTEGER PRIMARY KEY,
 dc1 DECIMAL(12, 4),
 dc2 DECIMAL(30, 30),
 dc3 DECIMAL(65, 0),
 dc4 DECIMAL(65, 30));
INSERT INTO t VALUES
(0, -99999999.9999, -.999999999999999999999999999999,
    -99999999999999999999999999999999999999999999999999999999999999999,
    -99999999999999999999999999999999999.999999999999999999999999999999),
(1, 0, 0, 0, 0),
(2, 99999999.9999, .999999999999999999999999999999,
    99999999999999999999999999999999999999999999999999999999999999999,
    99999999999999999999999999999999999.999999999999999999999999999999);

ALTER TABLE t SET TIFLASH REPLICA 1;
set @@session.tidb_isolation_read_engines="tiflash";
set tidb_enforce_mpp=1;
select dc1 + dc3 from t;
select dc1 - dc3 from t;

2. What did you expect to see? (Required)

3. What did you see instead (Required)

4. What is your TiFlash version? (Required)

tiflash-v5.3.0

Start by running the provided CREATE TABLE, INSERT, TiFlash replica, and dc1 +/- dc3 queries against TiFlash v5.3.0, then compare the results with the expected and observed screenshots. Done means decimal addition and subtraction no longer misbehave or overflow for the reproduced DECIMAL combinations.