Update on "[3/x] clean up casting functions: delete to_fp8_no_autograd"

Summary:

`ToFloat8ConstrFunc` was just calling `to_fp8_no_autograd`,
unify them to reduce confusion.

We can rename the function in a future PR, keeping PRs small for now.

Test Plan:

```
./test/test_everything.sh
```

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: [D60292694](https://our.internmc.facebook.com/intern/diff/D60292694)

[ghstack-poisoned]

Author: vkuzo

float8_experimental/float8_tensor.py

@@ -169,7 +169,7 @@ def forward(
 
         if isinstance(bits_fp8, DTensor):
             assert isinstance(
-                x, DTensor
+                scale, DTensor
             ), "Expected Float8 scale to be a DTensor if bits_fp8 is a DTensor"
             bits_mesh = bits_fp8.device_mesh
             bits_placements = bits_fp8.placements