As you will be unsurprised to learn, this is a common error for students learning about proofs. Two authors who write about this topic are A and J Selden; here is a citation and excerpt (pdf 472/598):
Selden, A., & Selden, J. (1987, July). Errors and misconceptions in college level theorem proving. In Proceedings of the second international seminar on misconceptions and educational strategies in science and mathematics (Vol. 3, pp. 457-470). Link.
(Side-note: It seems that "if it rains" is a common example! I see AndrewC also includes the statement: If it's raining, there are clouds in the sky. See also my example regarding the statement I'll go unless it rains as posted on MESE here.)
I have heard, in the past, instructors for proof-writing courses emphasize the last point about mathematical definitions using the if convention when they really mean if and only if. Moreover, this if vs. iff problem is mentioned by mweiss, who notes that many of the "if, then" propositions in eighth grade geometry are, in fact, biconditional. So: We have here one suggested answer, which is that you emphasize when the word "if" is used to mean "if, then" and when it is used to mean "if and only if"; perhaps you give examples in which you ask students more often to prove a statement, and then to figure out whether or not it is biconditional.
Toscho's comment gives a nice example: If the polygon $S$ is a square, then the polygon $S$ is a rectangle. Prove the statement. Is the converse true, too? Why or why not?
Looking at your three points, I have now gone beyond 1 (by using the word "converse").
You mention that you would like to teach somewhere between 2 and 3. I would actually object to the idea of 2, insofar as the word "opposite" can have different meanings. In this particular context, you give the example of $+$ and $-$ as being "opposites" because they are "inverses." Of course, in propositional logic, the implication $P \implies Q$ has both a "converse" (the subject of your question here) and an "inverse." Which one is the "opposite"?
There are other subtle issues with figuring out the converse of a proposition. For an example from group theory, see:
Hazzan, O., & Leron, U. (1996). Students' use and misuse of mathematical theorems: The case of Lagrange's theorem. For the Learning of Mathematics, 16, 23-26. Link.
in which the authors write:
The phenomenon of experts vs. novices relating to mathematical statements/problems at different levels (i.e., on a structural vs. superficial level, respectively) is recurrent. For another example of this and related citations, I refer you to my MESE response here.
As for level 3: I think this is the best choice. I am quite confident that you can intersperse real-life examples among geometric examples and still keep everything digestable for eighth grade students. (Incidentally, when I took geometry as a ninth grader, we began by covering propositional logic: implications, negations, and/or, truth tables, etc.)
Summary: Teach students that "if $P$, then $Q$" statements have as their converse "if $Q$, then $P$". Oftentimes, a true "if, then" statement run into by students will have its converse be true as well; however, this is not always the case. To help students understand this point, introduce the word converse and repeatedly ask about "if, then" statements both what their converse would be, and whether the converse is true or false.
Finally: An oft-incorporated technique in proof-writing classes is to show students examples of incorrect proofs. (See, e.g., Brendan Sullivan's MESE response here.) If you are observing a particular error frequently (e.g., students using a proposition's converse erroneously) then try writing up sample "proofs" with this mistake and ask students to critique them.