Better handling of RBend entrance and exit face angles

[cr-xu]

@cr-xu we are struggling with this PR a bit and I think you might have useful input.

In RBend there might be an issue with e1 and e2. If I instantiate an RBend and then access my_rbend.e1, the e1 I get is different from the one I set in the instantiation, by being in "Dipole space" rather than "RBend space". I believe that's an issue.

@Hespe and I were brainstorming solutions and one idea was to define two different e1 and e2 ... something like dipole_e1 and rbend_e1, where rbend_e1 would just be a computed property of the RBend class that computes based on dipole_e1.

What are your thoughts on this? Are you aware of any conventions already present in other codes or in general notation?

Originally posted by @jank324 in #289 (comment)

• The RBend should have new properties that agree with the values set by the users to avoid confusion. (Solved by Add clone method for beams and elements #289)
• Better documentation showing the difference in e1 and e2 for Dipole (SBend) and RBend

[cr-xu]

For example, a similar treatment is done in bmad (c.f. bmad manual Chapter 4.5).