For me the thing that's been a game-changer in recent years is the "Cracking the Code" series/curriculum. It reads like an info-mercial when summarized, you keep expecting someone to say "this one weird trick pro shredders don't want you to know about helps them play scale runs you can only dream about," except Troy Grady's approach is extremely logical and thoughtful, extremely well researched, and makes intuitive sense.
The short version is basically this - the hard thing about alternate picking isn't picking along a single string, which most people can do quite quickly with little practice at all, but it's
switching strings, and the reason that's hard is you have to lift the pick over the strings or get it over and to the other side of the string somehow, in a mechanically efficient manner, which is easy to do at slow speed, but extremely challenging and inefficient at high speeds. Pretty much any guitarist who can rip out fas scale runs has found SOME way to do this, but as a Yngwie fan the first time he was able to figure out how a world class player was doing this was watching slow motion footage of Yngwie's picking hand and realizing his pick stroke was angled, so he would bury downstrokes, but upstrokes would naturally lift away from the guitar making it easy to change strings after an upstroke, and every single fast "signature lick" of Yngwie's either changed strings on an upstroke, or on a downstroke sweeping to the next thinnest string, or would just omit one picked note and use a legato note instead to facilitate the change, and Yngwie had built his entire style around simply never changing strings after a downstroke and moving to a thicker string.
This is worth watching anyway just because it's pretty cool material, and brings back a lot of the feeling of sitting in your bedroom and having your mind blown by someone doing something you couldn't even imagine on a guitar, the first time you hear them. This was an early project and they've abandoned a lot of thew terminology as their understanding has changed a bit and as they've seen where people tend to get cnfused a little (Yngwie is now an "escaped upstroke or "USX" player, rather than a "downward pickslanting" player, since the actual slant of the pick has way less to do with it than the pick's trajectory) and I actually don't pick anything like this at all (I'm primarily escaped downstrokes, but with a rotational double-escape motion I'll use to facilitate certain transitions, not unlike a slower and sloppier Michael Angio Batio), but it's a pretty good intro.
EDIT - I guess, just to expand on this, the "start slow, gradually increase metronome speed" is NOT a good way to develop fast picking-hand speed. It's useful for coordination, and I've been doing legato drills like this latehy where it's definitely helped me build better
fretting hand evenness and control, but as the traditional way of becoming a faster alternate picker, it's actually not very good. The problem is the very thing that makes being a high-level alternate picker so challenging - being able to get the pick up and over a string and on to te next one while switching strings - doesn't even become a factor at lower speeds where pretty much anything works, and paradoxically fast, sloppy practice is pretty much the only good way to iron out a picking mechanic that's physically capable of working at high speed, even if at first it's irregular or poorly controlled. Picking is a naturally rhythmic motion, control will come with time, but to learn how to solve these mechanical problems, you have to practice in an environment where the problems even
exist.