Support torch_fp8

[cyita]

Description

• Add torch_fp8, torch_fp8_e5m2 and torch_fp8_e4m3

model = optimize_model(model, low_bit="torch_fp8", torch_dtype=dtype)
model = model.to("xpu")
output_fp8 = model.linear(input)

1. Why the change?

2. User API changes

3. Summary of the change

4. How to test?

• N/A
• Unit test: Please manually trigger the PR Validation here by inputting the PR number (e.g., 1234). And paste your action link here once it has been successfully finished.
• Application test
• Document test
• ...

5. New dependencies

• New Python dependencies
  - Dependency1
  - Dependency2
  - ...
• New Java/Scala dependencies and their license
  - Dependency1 and license1
  - Dependency2 and license2
  - ...

[rnwang04]

LGTM.
There may need more restrictions for this new type, like only support Linux / only used for BMG / save & load etc. But these are not urgent, we can consider such issues later.

[qiyuangong]

LGTM

[cyita]

PR validation https://github.com/intel-analytics/ipex-llm-workflow/actions/runs/15440629456