Current code allows for IllegalArgumentException to be just logged at trace level, effectively swallowing the exception. This has lead to a number of hard-to-diagnose issues, such as YANGTOOLS-1104.

In order to improve diagnostics and fail-fast behavior, investigate the choice of IllegalArgumentException in SchemaAwareApplyOperation.from() and the policy of catching that exception in DataNodeContainerModificationStrategy.

We need to differentiate illegal state type of behavior (as evidenced in YANGTOOLS-1104, which is missing code) from illegal argument type of behavior (where the user did something wrong).