Several thoughts on the subject.
1. Commercial ammunition is not loaded with canister grade powders. The powder they use is made in large production lots to a wider general specification, and the actual load data is then developed in-house by a ballistician.
For example DuPont will make batches of colloidal ball powder, and these batches are then analyzed for their burn traits and then mixed to get the burn traits specified by the commercial customer, which are a bit broader than the specifications for canister grade powder.
As an example of this, the US military used WC846 for the .7.62mm M80 Ball cartridge. When the 5.56mm M193 round was in it's long and difficult development process, one end of this powder specification was used to load M193 ammunition. That end of the specification was eventually just re-designated as WC844, creating two narrower specifications that had previously been a single specification.
H335 is generally regarded as the canister equivalent of WC844, while BLC-2 is generally regarded as the canister equivalent of WC846. Now look at a hand loading powder burn rate chart and note how far apart H335 and BLC-2 are on that chart. The original WC846 specification covered that entire burn rate range.
Consequently, anytime you see someone quoting a specific charge weight for a military or commercial load, it's just a nominal charge weight, not an actual weight that will give you identical performance with a canister grade powder.
2. Canister grade powders sold for the hand loading market are held to much narrower specifications as the average hand loader does not have the pressure test equipment to develop their own load data within specific pressure limits.
3. However, there is still variation from lot to lot. Win 296 and H-110 is a particularly interesting example. Back in the day before Hodgon came out and said that Win 296 and H110 were the same powder, you'd see both powders in the same reloading manual with significantly different charges. For example consider Hornady's 100 gr .30 M1 Carbine load in their 2003 sixth edition loading manual: 13.3 gr of H-110 versus 14.1 gr of Win 296 as the starting load. That's a 6% difference in what is supposedly the same powder, using the same bullet, case, primer, barrel, test equipment and test protocol.
Given that H-110 and Win 296 were the same powder at that point in time, that difference represents variation from lot to lot for the same canister grade powder.
Now in the distant past, it's possible that H-110 and Win 296 were more distinct, perhaps different ends of the same general burn rate specification. In any case, it needs to serve as a warning that there is variation even in canister grade powders and the 6% difference here why you see published advice to always start at least 10% below a max load, or to use caution with a max load when developing a load.
4. It's logical to assume the powder manufacturer knows the most about the powder, but that's only part of the load. If you ask a ballistician they will point out that the "stickiness" in the bore of projectiles of the same weight and caliber will vary. That is due to variation in bearing surface, in jacket alloy, harness and thickness, in the hardness of the core alloy, and in minor dimensional differences in the diameter of different bullets.
The end result is that when you see a powder company data for "125 gr" bullets for a .357 magnum load, it has to reflect the very broad range of available 125 gr .357 bullets and either the data will be adjusted accordingly, or it will be made clear that the hand loader needs to approach the limits with caution. You'll see the same thing in bullet company manuals like Hornady, where they may list several 125 gr .357 Magnum bullets above the load data. The odds are they have developed the data based on the "stickiest" bullet for the sake of safety, so in addition to powder variation from lot to lot, you'll get variation from bullet type to bullet type.
5. I'm not a fan of colloidal ball powders in a revolver for anything other than maximum loads with heavy bullets.
There is a myth flying around that the powder is completely burned before it leaves the case, but that doesn't seem to hold up in actual practice when you find un burned powder residue even with longer barrel pistols. And, even if the residue is burnt, it's still flowing through the forcing cone in a very hot state along with hot gas. The combination of hot gasses and gritty residue from ball powder creates a great mechanism to erode the forcing cone in a revolver.
I've also noted with the .357 Magnum that ball powders also have a tendency to blow particles back at me, and those particles are obviously being blown through the cylinder gap and being bounded off the frame or forcing cone and/or being blown by the pressure coming through that gap to strike me in the face.
I don't get that with a powder like Unique, where I'll only lose around 100 fps even in a 6" barrel compared 2400 or Win 296. Even better, I can actually get about 100 fps more velocity in a 2 3/4" or 3" barrel in the .357 Mag with Unique than I can get with 2400 of Win 296. (Before someone mentions the general advice that slow powders produce more velocity in the .357 magnum regardless of barrel length, be advised I'm basing my observations on extensive personal load development with a chronograph. I tend to put a lot more faith in my observed results than in what I read on the internet, or on general "wisdom".)
6. Erosion is reported to be greater with bullets of 125 grs or less bullets than bullets in the 135-158 gr range. The theory here is that the shorter bullets leave the case before the bullet has actually engaged the forcing cone, allowing the hot gas to race around the bullet and pre-heat the forcing cone before the bullet arrives, thus creating more stress on the forcing cone when the bullet arrives.
I'm skeptical for three reasons.
First, in terms of thermodynamics the time available between gas arrival and bullet arrival at the forcing cone is so brief that there is no time for the hot gas to transfer enough heat to the forcing cone to make any difference.
Second, the only prohibition I've ever read from a revolver manufacturer on the subject has been S&Ws caution about using 110 gr bullets in it's titanium cylinder revolvers due to erosion concerns.
Third, the observed phenomenon of cracked and eroded forcing cones arose with the change in law enforcement training practices with the .357 Magnum. Prior to about 1970, pretty much every department using the .357 Magnum practiced with .38 Special ammunition and reserved .357 Magnum for duty use only. After some law suits accusing law enforcement agencies of under training police officers by using .38 Special rather than .357 Magnum duty loads, most departments switched to .357 Magnum for training. With a steady diet of .357 Magnum loads, revolvers like the Model 19 started having issue with cracked forcing cones. That wasn't a surprise as the Model 19 was designed to be shot primarily with .38 Special with .357 Magnum intended to be shot much less frequently - on a ratio of perhaps 1 in 100.
However at this same time, .357 Magnum 125 gr loads became the overwhelming favorites with law enforcement agencies and the 125 gr bullet thus got the blame when problems started to arise.
7. I do feel colloidal ball powders do have a valuable place in .357 Magnum loads, but I reserve them for maximum loads where no other powder will suffice, and I use them with the full understanding that a steady diet of those loads will shorten the life of my revolver.
Now for most folks it probably doesn't make much difference, as they don't put enough rounds down range for the reduced life to ever pose a problem. But if you're one of those people averaging 100-200 rounds or more a week, it's going to be an issue and you'd be better served by a different type of powder.