release-22.2: opt: fix ambiguous column error for computed column expressions

[blathers-crl]

Backport 1/1 commits from #104924 on behalf of @mgartner.

/cc @cockroachdb/release

---

This commit fixes a rare bug that caused "ambiguous column" errors for
INSERT .. ON CONFLICT .. DO UPDATE statements.

Consider the schema and statement:

CREATE TABLE t (
  id INT PRIMARY KEY,
  a INT DEFAULT 1,
  b INT,
  c INT AS (a + b) STORED
)

INSERT INTO t(id) VALUES (0) ON CONFLICT (id) DO UPDATE SET b = 1

In the SET clause, an explicit reference to a is ambiguous. It could
be a reference to t.a, the a of the conflicting row(s), or it could
be a reference to excluded.a, the a of the insert row(s) from the
VALUES clause. If the SET clause contained b = a, the result would
be an "ambiguous column" error, which is correct and matches Postgres's
behavior.

With the SET clause as b = 1, there should be no ambiguity. However,
prior to this commit, ambiguity arose when building an implicit SET
expression to generate the new value for the computed column c. When
building a + b, we used the same scope used to build the b = 1
projection: the scope of the SET clause. Using this scope makes a
ambiguous, even though it always refers to t.a - the a of the
conflicting row(s).

Normally, this ambiguity is avoided by calling
mutationBuilder.disambiguateColumns, which is called before computed
column expressions are built. However, it's not effective in this
specific edge case where the new value of b and the default value of
a are the same expression, 1. In this case, we conserve column IDs
by using the same ID for both, which prevents
mutationBuilder.disambiguateColumns from removing the ambiguity in the
SET clause scope.

The bug has been fixed by creating a new scope for resolving column
references in computed column expressions. The scope contains no columns
with duplicate names, so ambiguity is impossible.

I believe the long-term fix is to avoid using scopes when building these
synthesized expressions. I attempted to do this, but there are many
tricky edge cases to iron out and the change quickly grew beyond the
size of a comfortable backport. This future work is tracked by #61298.

Fixes #102909

Release note (bug fix): A bug has been fixed that caused
INSERT .. ON CONFLICT .. DO UPDATE queries to incorrectly result in an
"ambiguous column" error. The bug only presented if the target table had
a computed column with an expression referencing a column with a DEFAULT
value.

---

Release justification: Fix for a bug that causes ambiguous column reference
errors for some INSERTs.

[blathers-crl]

Thanks for opening a backport.

Please check the backport criteria before merging:

• Patches should only be created for serious issues or test-only changes.
• Patches should not break backwards-compatibility.
• Patches should change as little code as possible.
• Patches should not change on-disk formats or node communication protocols.
• Patches should not add new functionality.
• Patches must not add, edit, or otherwise modify cluster versions; or add version gates.

If some of the basic criteria cannot be satisfied, ensure that the exceptional criteria are satisfied within.

• There is a high priority need for the functionality that cannot wait until the next release and is difficult to address in another way.
• The new functionality is additive-only and only runs for clusters which have specifically “opted in” to it (e.g. by a cluster setting).
• New code is protected by a conditional check that is trivial to verify and ensures that it only runs for opt-in clusters.
• The PM and TL on the team that owns the changed code have signed off that the change obeys the above rules.

Add a brief release justification to the body of your PR to justify this backport.

Some other things to consider:

• What did we do to ensure that a user that doesn’t know & care about this backport, has no idea that it happened?
• Will this work in a cluster of mixed patch versions? Did we test that?
• If a user upgrades a patch version, uses this feature, and then downgrades, what happens?

[cockroach-teamcity]

This change is 

[rharding6373]

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @cucaroach, @mgartner, and @msirek)

[mgartner]

Probably worth letting this bake for a bit.