feature(core) Stricter validation for atomic values

[jeromesimeon]

Signed-off-by: Jerome Simeon jeromesimeon@me.com

Issue #157 & #158

A stricter validation/serialization semantics for atomic types and values in CTO models.

Changes

• Request CTO validation is not working #157 Atomic values in the JSON have to be strictly of the type declared in the model. E.g., true validates against Boolean but "true" does not. E.g., 3.14 validates against Double but does not validate against Integer or Long. E.g., "1" validates against String but 1 does not.
• Serialization of Infinity is null #158 Serializing non-finite Doubles (NaN or Infinity or -Infinity) throws a validation error
• Updates the tests accordingly

Flags

• This is a breaking change
• The PR is against a new release-0.83 branch

[dselman]

We should add removing the "ergo" stuff that snuck into Concerto v1 imo...

[jeromesimeon]

We should add removing the "ergo" stuff that snuck into Concerto v1 imo...

We should definitely discuss how we want that to work, from a code-base / feature perspective!

[jeromesimeon]

We should add removing the "ergo" stuff that snuck into Concerto v1 imo...

This is somewhat orthogonal to this specific PR and issue. I opened a new issue to track the "ergo stuff removal" here: #162