sql/parser: Permit placeholders for OID values

[nvanbenschoten]

Part of #14245.

Previously the tOID type claimed that it was ambiguous through its
IsAmbiguous method. This meant that a placeholderTypeAmbiguityError
would be thrown when issuing queries like SELECT 1::oid = $1. This was
incorrect, as an OID type is fully defined.

This change fixes this and also improves the tOid type slightly. It
gets rid of its redundant typeName field. Additionally, it improves
the type's Equivalent and FamilyEqual methods, which would have
given us issues if one of the "reg oid" variants (typeRegClass,
typeRegProc, etc.) ever entered our type system.

---

This change is 

[tamird]

Reviewed 2 of 2 files at r1.
Review status: all files reviewed at latest revision, 1 unresolved discussion, some commit checks pending.

---

pkg/sql/parser/type.go, line 562 at r1 (raw file):

		return true
	}
	_, ok := UnwrapType(other).(tOid)

this can delegate to FamilyEqual, no?

---

Comments from Reviewable

[jordanlewis]

Thanks! I was working on debugging the problem in #14245 this weekend as well, and I also noticed and corrected (but didn't commit) the isAmbiguous implementation. Unfortunately, this doesn't fix the problem I noticed, despite these tests passing. I'll put more details in #14245, but for now I think we'll need to remove the Fixes statement from this PR.

---

Review status: all files reviewed at latest revision, 1 unresolved discussion, some commit checks pending.

---

Comments from Reviewable

[nvanbenschoten]

TFTRs! Removed the "Fixes" wording from the commit message.

---

Review status: 1 of 2 files reviewed at latest revision, 1 unresolved discussion.

---

pkg/sql/parser/type.go, line 562 at r1 (raw file):

Previously, tamird (Tamir Duberstein) wrote…

this can delegate to FamilyEqual, no?

Done.

---

Comments from Reviewable

[tamird]

Reviewed 1 of 1 files at r2.
Review status: all files reviewed at latest revision, all discussions resolved, some commit checks pending.

---

Comments from Reviewable