Investigate other uses of `serde(untagged)` for performance problems

[cdisselkoen]

Describe the improvement you'd like to request

In #1308 we fixed a performance issue (#1284) by replacing an automatically derived serde(untagged) deserializer with a custom one, achieving orders of magnitude speedup on a pathological case. We should investigate if it's worth applying the same pattern in other places.

Describe alternatives you've considered

No response

Additional context

No response

Is this something that you'd be interested in working on?

• [ ] 👋 I may be able to implement this internal improvement
• [ ] ⚠️ This feature might incur a breaking change

[john-h-kastner-aws]

Fortunately, the list is not too long. Only one is a recursive data structure, which I believe was key in the pathological performance issue

• [x] EST Expr fixed by #1308
• [x] JSON schema Type does not derive Deserialize after #132, which essentially applied the same fix
• [ ] CedarValueJson. Recursive data structure. I suspect this means it is subject to the same sort of pathological inputs.

All the rest are worth taking a closer look at, but they are not defined recursively, so the cost of the backtracking parsing for the derived untagged should only be paid once.

• [ ] EntityUidJson
• [ ] ExtnValueJson
• [ ] EST ActionInConstraint
• [ ] EST PrincipalOrResourceInConstraint
• [ ] EST EqConstraint
• [ ] FFI Policy
• [ ] FFI TEmplate
• [ ] FFI StaticPolicySet
• [ ] FFI Schema