[llvm] [AArch64] Disconnect FeatureUseScalarIncVL from FeatureSVE2. (PR #114568)

[Sander de Smalen via llvm-commits]

================
@@ -797,6 +794,9 @@ def FeatureUseFixedOverScalableIfEqualCost: SubtargetFeature<"use-fixed-over-sca
   "UseFixedOverScalableIfEqualCost", "true",
   "Prefer fixed width loop vectorization over scalable if the cost-model assigns equal costs">;
 
+def FeatureDontUseScalarIncVL : SubtargetFeature<"dont-use-scalar-inc-vl",
----------------
sdesmalen-arm wrote:

I found the double negative in `-mattr=-dont-use-scalar-inc-vl` confusing and the current logic is a bit simpler (although I'm not particularly happy with the explicit partial string matching, which I guess is the downside of doing it this way). A name change would avoid the double negative, but thinking about this; why don't we just make it an LLVM option rather than a feature? 

That way, we can do something like:
```
if (ForceScalarIncVL.getNumOccurrences())
  UseScalarIncVL = ForceScalarIncVL.getValue();
else if (hasSVE2() || hasSME())
  UseScalarIncVL = true;
// else UseScalarIncVL = false; (would be default)
```

We can still make use of the `UseScalarIncVL`  Predicate in TableGen, it just wouldn't be defined through a target feature.

https://github.com/llvm/llvm-project/pull/114568