Reading tonight's goings on, I'm not sure if these comments are helpful, but I hope they are.
Metering technique for negative film is quite different to transparencies, and digital is slightly different again (but similar to trannies). With negs you can adjust the final tone range in the print, which makes a fundamental difference. You can't with trannies. With digital there is some scope to adjust the final result if you shoot Raw, which is a bit of highlight headroom which can be exploited with the 'expose to the right' technique. (JPEGs are just like transparency film exposure-wise.)
Different films have different dynamic range. 7 stops is a rule of thumb. Generally, the higher the ISO, the shorter the range, and it's the shadow end which suffers.
Spot metering is generally most appropriate to shooting negs, in two ways: 1) point it at an area and it will translate the reading into an exposure that will nail that tone to 18% grey on the film. With neg film, this will give you maximum exposure latitude either side, which will then give you an optimum print. This is hopeless for tranny film or digital, unless the area you've chosen happens to be 18% grey. 2) Choose the brightest important highlight area, and the darkest important shadow, and the spot meter will tell you the range between them. If it's within 7-stops, you're okay. Set your exposure mid-way and you should get the lot. If it's outside 7-stops, you've got to choose which way to compromise.
A shorthand version of 2) is to spot read the important highlight, and just add two stops. This usually works (with all film and digital) because highlights are more important than shadows and if you've got the brightest area pinned to the top of the tone curve, everything else will just hang off it.
Shooting transparencies or digital JPEG, an incident reading is the best, and there is little room for manoeuvre - maybe half a stop according to preference. Digital Raw has some highlight headroom, about one stop, according to camera model.
The spot meter doesn't know if the tone it's seeing is light or dark, only you know that. And using experience, you know where you want that brightness area to sit on the tone curve.
This is quite in-depth theory. Understanding is important, but due to inherant inaccuracies in the system, it is extremely difficult to operate with any dgreee of precision greater than plus/minus 25% at best. There is only ever one optimum exposure setting (depending on recording medium) and it makes no difference how this is calculated.
The example of the grad is a good one. Yes, you would effectively meter the sky separately and depending how bright it was above the level of the main scene, decide what strength of grad to use. And then you can vary the effect of the grad with f/number also (higher f/number pulls the dark line down quicker and harder).