Add tuned moe config for qwen1.5_moe_A2.7B

[sky0530]

This PR add tuned qwen1.5_moe_A2.7B moe config with tp = 1, batch = 1

[sky0530]

Add more batch size support.
This PR add tuned qwen1.5_moe_A2.7B moe config with tp = 1.
It's a small model. And it would not use multiple GPU to do inference

[divakar-amd]

Hi, thanks for the tuning configs. We would need a small update in the file names:

• Can we remove the trailing "_OAM" from the names of these 4 json files? We want to keep the names consistent across all the mi300 machines. (PR for more details)

Also, it will be good to see some perf gains numbers obtained after the tuning as a measure of correct tuning process.

[sky0530]

With tuned moe config, the Qwen1.5_moe_A2.7B can improve E2E 3.5% performance

[sky0530]

@divakar-amd I have changed the filenames. Plz help to review the PR

[divakar-amd]

LGTM. Approving!