Interesting point. To me, ISO control = dynamic range/image quality control.
That's not what the ISO control is intended to do, that's working with the side effects built into the ISO control, which vary from camera to camera.
Yes, definitely and unfortunately, it's camera-specific. I know it's a side effect of how ISO setting affects the reading and processing chain in camera, but it works.
There are lots of techniques that work due to exploitation of side effects. All of them require good knowledge of what those side effects are.
That's what I call 'ISO first' technique, and it has its problems. Essentially you are making a guess at the exposure you'll use in advance of metering it.
In many scenarios, acceptable shutter speed may vary from say 1/1000 to 1s, basically, 10 stops easily. There's a lot of room to adjust the exposure, and no need to even guess it.
You're still making a guess. In that case you want the slowest of that range of shutter speeds that, along with your choice of f-number and the scene illuminance, results in the largest exposure that your camera will accept. I think that for general landscape use 1s is pushing it. You lose leaf detail, unless the day is dead calm or the scene is barren.
Still the principle there is that you'll be working at the lowest ISO you have, but you've still made a guess that the scene will allow that. The thing about the exposure-first method is that it always works, you don't need an exception for this kind of shot or that kind of shot. Of course, when you get experience with different types of shot you can adopt shortcuts, as you do with most techniques.
The assumption that you make setting the lowest ISO for landscape shots is that none of your creative constraints will lead to a lower exposure than the nominal for the base ISO.
That's true. In some cases I need a specific shutter speed, e.g. around 0.5s for seascapes/waves/surf. But in such cases exposure gets to high and requires ND filters, as below.
Sure, in that specific case, you're always working with the limit that your camera will accept.
The using of base ISO and ND filters is exactly equivalent in exposure terms to stopping down.
ISO first works OK if you never leave base ISO, but if you do, generally has the effect of you ending up with a lower exposure than your creative constraints would allow, because you end up using the nominal exposure for the ISO that you guessed in the first place.
No, as above, in such cases I set the exposure first and then tweak ISO to get an ETTR-like histogram. So it's very close to Manual+Auto ISO, only I think I get more optimal ISO in the end. In such cases, t resulting ISO setting is also an indicator of prospective noise/IQ.
I don't think anyone is suggesting that you can't get 'more optimal' ISO if you set it yourself, just that the difference is small enough, and the convenience great enough to make auto ISO a practically better option. The general rule for optimising ISO after setting exposure is to set it as high as you can without clipping highlights. Auto ISO will do that, possibly requiring some EC to protect the highlights. Fine tuning involves stopping raising the ISO after the camera reaches the invariant point, which can be dealt with in many auto ISO implementations by setting an ISO limit, and making an alternate choice at the gain switch point if you have dual conversion gain. As I said before, I don't think that it's clear that you gain by moving to a higher ISO, and with auto ISO suitably set up, if that higher ISO doesn't clip the highlights it will be selected in any case.
My preferred technique is what I call 'exposure first', where you set the f-number for the minimum DOF you want, and the shutter speed for the motion blur you want (or don't want). Then you set the ISO to match. If you go down to the lowest ISO and the exposure is too big, then you reduce it. That is a straightforward and mechanistic procedure. It is easily automated (with auto ISO currently) and always leads to the best IQ and DR that your creative constraints allow.
Auto ISO uses in-camera metering with a risk of blowing the highlights. Normally I don't want to blow any highlights in a landscape shot (not talking about the sun in the frame), so I use the histogram, which effectively serves as a manual metering mode - again very close to auto ISO, as you described, only replacing auto metering with manual by histogram.
Yes, as above, there are places where you can make a better choice than auto ISO, but if you want to protect the highlights you can use highlight metering (with auto ISO) or an appropriate amount of EC. Histograms are not a universal recipe for avoiding blowing highlights. For a start, if you're using a raw workflow a JPEG histogram will give false alarms, and also the resolution of the histogram can be quite coarse, and there can be highlights which don't represent a large enough proportion of pixels to show up on the histogram.
Having the ISO setting may also be important in production to have consistent settings across all cameras used in the field.
Which settings are you talking about? If you want consistent settings, why not just set them consistently?
ISO is a reference point to synchronise different cameras so that they produce consistent output for future processing.
I don't think that is true. I'm trying to work out precisely what you mean by it. Possibly some more detail as to what you mean would help.
Especially relevant in video production. If a camera manufacturer completely removes ISO setting, as you suggest above, how would a video or stills camera operator make sure their settings match other cameras?
As I said, which settings are you talking about, and how would they 'match' or not 'match'? And how does ISO help them match. In the case of exposure, all you need is to know that the exposure is the same, but that will give a different effect on different cameras according to sensor efficiency and size. Also, ISO defines the lightness for the exposure at only one reference point (18% grey), so different cameras my be set to the same ISO, and conform to the standard, but produce output that looks completely different in both tonal range and colour.
, I'd be having a minor brain meltdown 
