opt: avoid planning uniqueness checks for upsert if new values are null

[rytaft]

We currently plan some unnecessary uniqueness checks on UNIQUE WITHOUT INDEX columns when planning UPSERT queries. There are a couple of reasons for this:

1. The input for an UPSERT may need to be wrapped in an EnsureUpsertDistinctOn and a LeftJoin. Since the function OutputColumnIsAlwaysNull doesn't recurse into JOINs or GROUP BYs, we may not be able to determine that a particular column is always NULL. We'll need to update this function to add support for JOINs and GROUP BYs. This will fix the case where the insert and update mapping for a column are the same (both NULL).
2. For columns that have different insert and update mappings, we don't need to plan checks if (a) the insert mapping for at least one of the columns in the unique constraint is always NULL, and (b) none of the columns in the unique constraint are being updated (i.e., they don't have an update mapping).

Epic CRDB-13964

Jira issue: CRDB-3411

[github-actions]

We have marked this issue as stale because it has been inactive for
18 months. If this issue is still relevant, removing the stale label
or adding a comment will keep it active. Otherwise, we'll close it in
10 days to keep the issue queue tidy. Thank you for your contribution
to CockroachDB!