I put the missing link in. Not sure how I missed doing that, but I did.
Well, I'm showing my age. It's been 40 years since I did any chromium plating and that was in a lab environment, but I remember being frustrated by the hexavalent chromium and sulfuric acid solution etching steel faster than I could plate. At that point in time the only difference between the decorative and industrial hard chrome plating was not the plating solution but higher plating temperatures and current density for the latter to get thicker layers. Today they seem to use less toxic trivalent solutions for decorative plating. When I looked at an old lab notebook I found I could not meet the normal current density of hard chrome plating with the equipment I had, and that's the part I had forgotten. So you are correct that no nickel intermediate coating is involved.
A quick perusal of military plating specs shows the decorative type is class 1 and the hard chrome type without underplating is class 2. From QQ-C-320B (1974, last version before SAE-AMS-QQ-C-320 replaced it):
3.4.2.1 Thickness. The minimum, maximum or range of thickness for class 2 plating shall be as specified in the contract, purchase order or on the applicable drawing (see 6.2). If a thickness is not specified, the minimum thickness for the finished part shall be 0.002 inch or 2 mils (51 μm).…
So that may be one reason for the 2 mil number. I don't have a barrel drawing to verify it. I did run into one thread on THR (sorry, I didn't copy the link, but you can search it out if you are interested) in which it was mentioned that the barrel plating was 0.002" thick and that Armalite had to lap bores before applying it. (That makes sense as the hex chrome plating solution has poor throwing power and an uneven surface will get an unevenly thick plate. You have to form the lead anodes to mirror the surface you want to plate evenly, I recall.) But there are a couple of other reasons to think 0.002" is probably right.
It turns out, in plating big gun barrels, the process is more elaborate. The bore is first electropolished to remove microscopic unevenness, then is reverse current etched to activate the surface before the plating is actually applied. For the big guns the chromium plating is 0.002"-0.006" thick. That's in a table on page 30 of
this document. The document is about finding alternatives to electroplated chromium and is an interesting read because it covers the problems with plated bores and has a flow chart of the plating process early on and one for barrel making that is in the slides that start about half way through the document.
The last reason to think 0.002" chrome plating is minimum, even in small arms barrels, is mentioned in the text of that last document. Part of the protection mechanism provided by the chromium is not mechanical wear resistance. It is providing a thermal barrier to exposure of the steel substrate to the high side of the temperature gradient that develops during firing and which penetrates about two thousandths of an inch into the bore. (I got that from a study a couple of years ago; if I can find the link, I will come back and put it in). When the steel is exposed directly to that transient gradient, it detempers the surface martensite, weakening it, and that is part of the mechanism of forming the alligator skin heat stress cracking you see in bore throats.
Another aspect of heat stress cracking is the differential expansion of the steel at the surface side of the gradient that finally breaks those alligator skin squares off. Chromium has only half the linear temperature coefficient of expansion of steel, so it doesn't endure as much stress from the heat exposure as steel does. In other words, the chromium is acting as a more temperature tolerant thermal barrier as well as being a wear barrier. Since cannon and small arms are exposed to similar powder combustion temperatures, it makes sense to suppose the minimum coating thickness of chromium will be of similar magnitude to that of the big guns. While a thin plating of chromium can get you the wear resistance, it will be insufficient to achieve the thermal protection.