Clarify restrictions on steady state

[jjhugues]

In connection to osate/osate2#2546

[jjhugues]

@lwrage Can you please propose a resolution that corresponds to the test you implemented?

[lwrage]

OSATE now supports steady state error transitions given as

• transitions where the target error state is given as same state
• transitions where the target error state name is the same as the name of the source state.

As it is currently implied that both ways of specifying a steady state transitions are legal it is not strictly necessary to change the annex. However, it would be helpful to explicitly state that both ways are legal and that same state can additionally be used as a shorthand to avoid having to write multiple steady state transitions:

• When the source of a transition is all
• When the source is the name of a typed error state with more than a single error type as the type constraint.

[jjhugues]

Thanks; captured in the document with marker "Issue #51"