While a bot might try to follow it (because it is, in it simplest form, a type of link), that will not in any way adversely affect you.  That tel: in the tag will tip them off that it is a telephone number and/or should be click-to-call. So no link equity will be lost, you won't start seeing tons of 404 warnings, or anything of that sort.

Are the categories helpful for the customer? On one hand you don't want to lump too many things into one category when they can be broken out into more granular categories that better serve visitors. On the other hand, it won't help you or your customers if you get too granular and break everything out into categories based on the mot insignificant details.

While keyword cannibalization is a concern, serving your visitors/customers what they want and how they prefer to see it will likely improve metrics more on your site than concerning yourself with a nebulous concept like "how many categories is too many." If you have 200 different categories but they are well targeted and you want to add another (or ten more) that are also equally well targeted, then why wouldn't you do it?

Personally I think you should set up a process whereby every time a vehicle and/or part is removed, you have someone automatically 301 it to the previous step in the site navigation. So when "blue widget 3" is removed from the site, anyone landing on that page or who has it bookmarked winds up on the "Widget" category page. Now there may not be an easy way to do it right this second because of how many there are now, but if you get in the habit of doing it and slowly work toward fixing the others then you'll be in a good position in the future to keep this from being an issue again.

Now if you really don't want to attempt that... 404s aren't necessarily horrible (too many can be). If your site is properly serving 404s then you won't be penalized for it but in this case you might want to consider using 410 status codes. Its a stronger signal for removal than a 404 and you don't plan on the product ever coming back so marking it Gone should get it removed from the index faster while also helping to keep you from competing against yourself in the SERPs when a new but similar product comes into stock.

Have you double-checked your rankings either through another tool or through unpersonalized searches to see if the Moz data is correct? I have a few search terms on two of my campaigns that regularly swing from 1st one week to "Not in top 50" the next according to Moz when in reality they barely shift ever.

If a URL was indexed and has since had a canonical added to it pointing to another page, it will eventually disappear from results. Basically the pages gets consolidated with its canon page. If the bots choose to respect the canonical tag in that instance, all signals get passed to the canon page while still allowing the page and information to be accessible by human visitors. As such, there's no reason to keep the page in the index because you're telling the bots that another page is the correct page instead. This is not the same as NoIndexing a page but will eventually remove a page from the index much in the same way that a 301 will pass equity along to another page while eventually removing the redirected page from the index in favor of the page being redirected to.

Not exactly. Its not so much that the canonical "supersedes" an index, follow tag.... a canonical tag establishes equivalency while a NoIndex is more like a "does not equal." The Index, Follow is still there and being seen by bots as they crawl... in fact, if you had NoIndex on a page with a Canonical Tag, it may not even see the canonical at all since you told it to NoIndex the page. The Meta Robots Index tag comes first allowing the bots to crawl and index the page but then the canonical sets up equivalency to a separate page. So if your canonical tag is being respected, it doesn't wind up doing the same thing as a NoIndex (though it may seem that way) nor does it do the same thing as a 301 (though there are similarities in how equity is passed). Since a canonical establishes an equivalency, you'll find that the Canon Page will eventually take the place of the Canonicalized Page in search results because you're telling them the Canonicalized Page _is _the Canon Page & that the Canon page is the right version of both.

From my understanding when using hreflang to denote English speakers in the entirety of the European Union you wouldn't need a regional tag on there.

I.E. if you targeting English speakers in Canada it would be "en-ca" whereas targeting English speakers in the UK would be "en-gb". But when targeting English speakers in all 27 (?) EU nations at once then you would just use "en".

You might be better served by using a canonical to point the parameters to the base page. I.E. /shoes?gender=1 with a rel="canonical" pointing at "/shoes". Depends on the variety of the content of the pages, if you're cannibalizing your own keywords, etc.