[Perf]: Support non-contiguous input for `dynamic_scaled_int8_quant` and `dynamic_per_token_scaled_fp8_quant`

[yewentao256]

🚀 The feature, motivation and pitch

From #19452

We temporally force input to be contiguous, which may cause a high copy overhead when input is not contiguous.

I'm trying to work on this to support non-contiguous input for the quant kernel.

Scope:

• dynamic_scaled_int8_quant
• dynamic_per_token_scaled_fp8_quant

Alternatives

No response

Additional context

No response

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.