release-21.2: optbuilder: do not create invalid casts when building COALESCE and IF

[mgartner]

Backport 1/1 commits from #77608.

/cc @cockroachdb/release

---

The optbuilder no longer creates invalid casts when building COALESCE
and IF expressions that have children with different types. Expressions
that previously caused internal errors now result in user-facing errors.
Both UNION and CASE expressions had similar bugs that were recently
fixed in #75219 and #76193.

This commit also updates the tree.ReType function to return ok=false
if there is no valid cast to re-type the expression to the given type.
This forces callers to explicitly deal with situations where re-typing
is not possible and it ensures that the function never creates invalid
casts. This will make it easier to track down future related bugs
because internal errors should originate from the call site of
tree.ReType rather than from logic further along in the optimization
process (in the case of #76807 the internal error originated from the
logical props builder when it attempted to lookup the volatility of the
invalid cast).

This commit also adds special logic to make casts from any tuple type to
types.AnyTuple valid immutable, implicit casts. Evaluation of these
casts are no-ops. Users cannot construct these casts, but they are built
by optbuilder in some cases.

Fixes #76807

Release justification: This is a low-risk change that fixes a minor bug.

Release note (bug fix): A bug has been fixed that caused internal errors
when COALESCE and IF expressions had inner expressions with different
types that could not be cast to a common type.

[cockroach-teamcity]

This change is 

[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?

[msirek]

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @rharding6373)

[rharding6373]

Reviewable status: complete! 2 of 0 LGTMs obtained