The Entity-Relationship Model proposes that reality be viewed from three points of view, namely:
a) the objects that make up reality,
b) the types of information or characteristics you want to know about the objects that make up reality and
c) how these objects interact with each other.
In this way, Entity-Relationship Model is composed of three concepts: Entity, Attribute and Relationship . The objects that make up reality are Entities. The characteristics that one wishes to know about the objects that compose the reality are the Attributes. The way objects interact with each other is the Relationship.
So your first two questions lead me to the same answer: Yes, both models are correct, the first one being the best. The more detail you add in a correct and coherent way to the model as taught by theory, the better the understanding / understanding / ease of development will be. Of course, any additional effort should match within a cost-time-result line, but I will not go into those details that run the gamut ...
As for the attribute , I learned that:
a) Identification : The identification of each attribute must be composed of a singular noun and, if necessary, one more qualifier that characterizes it. Capital letters must be used. For example: NAME, DESCRIPTION, DATE OF BIRTH
b) Definition : Each attribute should be given a brief description of its meaning to the Organization. In addition, the description of an attribute must be complemented with a local or global data type that characterizes it. For example: REGISTRATION NUMBER (INT, 05).
That is, if these attributes are with many characters it is because there is something that contradicts the item A of our theory. Therefore, it would be better to review their identification.