Before proposing reviewers, should I notify them?
I never asked them and never heard of anybody doing so. It’s the journal who is responsible to organise peer review, not you, and they have to and will humbly ask¹ potential reviewers anyway. By asking potential reviewers yourself, you cause the following problems:
Not every reviewer you suggest will be chosen by the editor. For a reviewer who will not be chosen to review, deciding upon your request will be a waste of time.
Even a reviewer who is eventually chosen will waste a little time by responding to the same request twice.
If a potential reviewer declines your request and you do not suggest them, the editor may still choose them on their own (most editors to whom I have talked make a point of choosing at least one reviewer who was not suggested). In that case, the potential reviewer may think that you ignored their wish.
Asking reviewers directly is a first step to breaching the independence of peer review as it opens up the path for manipulating peer review: “Sure I will review your paper [favourably] – provided that you will do the same thing for my next paper.” Some oppose even the mere option to suggest peer reviewers for this reason.
If you know only a limited list of potential reviewers for your paper, removing those who deny your request may make the list smaller than what is requested by the journal. If those reviewers only refuse when asked by the journal, you still did your duty of suggesting sufficiently many potential reviewers (and this may not even be a problem, if the journal requests a list that is longer than the number of actual reviewers).
You waste some of your time by asking the potential reviewers and waiting for their response may delay the actual submission.
If you show your paper to your potential reviewers privately, they have an advantage in the rare case that they try to steal your ideas or commit other dishonesties. By contrast, peer reviewers are often bound to terms of non-disclosure and similar and also the journal serves as an independent register of who had knowledge of the manuscript. This point does of course not apply, if you trust all potential reviewers or published a preprint of your paper anyway.
¹ Except perhaps for some journals with a questionably demanding demeanor.
You should never ask someone as an author whether they would be willing to serve as a referee for your paper. In addition to Wrzlprmft's issues:
It looks suspicious, like you may be filtering your suggested referees based on how enthusiastic they seem to you. This could happen even if it's honestly not your goal. For example, suppose you ask A, who replies "Wow, I can't wait to read it!", and B, who replies "I'm really busy but I could do it if necessary." If you suggest only A to the editor, are you biasing the process in favor of your paper, or just being respectful of B's time? There's no clear way to draw the line.
You're putting potential reviewers in an awkward position by making them tell you whether they could do it. For example, they may feel more comfortable telling an editor that they are too busy or feel the paper is outside their area of expertise, rather than telling you. They may also worry that giving you this information decreases their anonymity.
I would say that any contact between referees and authors should be minimized. Just mention any potential referees when submitting.
Secondly, my personal experience is, is that referees are chosen from the existing journal's database, regardless whether and how many potential referees are provided. This, on the basis of the types and quality of the referee's reports I received in a number of cases; for example quite poor and obviously not one of my suggested referees, or their response was way too knowledgeable about certain topics to meet the knowledge on those topics of my suggested referees.
And to add to that: I've talked to an editor-in-chief who gave a crash-course in article writing and he said that he would rather not use the potential referees as they are likely related to the manuscript author(s) in one way or another, professionally or otherwise. He told us that he, therefore, never provided any potential referees with his submissions, unless he had to.