Distinguish exceptions that occurs when handling message.

[taeguk]

We need to distinguish exceptions that occurs when handling message.
Now, there is a ambiguous division in our codes.
This means instability.
So, we need to divide those exceptions into four things.

1. Known exceptions occurs. And they mean a invalid message.

• Pass these exceptions. And pass a message.
• Pass these exceptions. And report a failure to sender.

2. Known exceptions occurs. And they just mean a known failure of functions.

They mean just "we can handle these exceptions"

• Handle these exceptions by known(=already defined) solution.

3. Unknown exceptions occurs. But we can recover a current state to be a state before occurrence of a exception.

They are likely to mean "bugs of codes" or a external problem. But we can recover a state.

• Recover a state. And ignore a message or report a failure to sender. And log about unknown exceptions for debugging.
• Die for immediate debugging.

4. Unknown exceptions occurs. And we can't recover...

They are likely to mean "bugs of codes" or a external problem. And we can't recover a state.

• Die

[taeguk]

I think it is related to some refactoring issues.

1. Documentation
2. A need of Module Root Exception.
   So, if we want to solve this issue, we need to solve some refactoring issues above.