experiment: repro and fix for issue #5504

[crusso]

(Implement Martin's revised deserialization for Candid options.)

This PR fixes coercion of values into nested optional types.

Before the fix from this PR, 5 : nat64 would coerce to opt 5 : opt na64 (as expected) and null : opt opt nat64 (unexpected). >This PR fixes the unexpected result to be opt opt 5 : opt opt nat64 (as expected).

Before the fix from this PR, 5 : int would coerce to null : opt opt nat. This PR fixes the result to be opt null : opt opt nat. The >motivation for this change is to enable using optional types as follows:

• outer option characterizes if a value (of any type) is provided;
• inner option expresses to optimistically deserialize the actual value into the given type (without failing to deserialize and without losing the information if a value of any type is provided).

This PR also adds multiple test cases covering the change to coercion (including coercion to nested optional types over null and reserved).

c.f.

• Rust fix fix: coercion of values into nested optional types candid#679
• Candid spec PR, in particular this commit
  dfinity/candid@90af305

[ ] TODO bump candid dependency to pick up revised test suite

[github-actions]

Comparing from bee6787 to 82ce6de:
The produced WebAssembly code seems to be completely unchanged.
In terms of gas, no changes are observed in 5 tests.
In terms of size, no changes are observed in 5 tests.

[mraszyk]

Would it be possible to make the tests more compact (and better readable) using generics (analogously to the Rust tests)?