[config] Add a llama2-7b-int8 test config for decode

[xy12181]

Also add a small accuracy check for the generate text prefix.

Checklist

Before submitting this PR, please make sure (put X in square brackets):

• I have performed a self-review of my code.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed.

[vipannalla]

looks good

[xy12181]

To the reviewers: below is my dev work flow.

• Fork the AI-Hypercomputer/maxtext main branch to my own maxtext repo;
• For a feature Update unit tests status badge in readme #1, I develop in my private local/feature1 branch;
• After it is ready for review, I cherry-pick the commit(s) into my public yong/dev1 branch. I only use this branch to submit my code for your review;
• I address reviewer comments in my local/feature1 branch, and then rebase it to yongx/dev1 branch and force upload -- I keep my yong/dev1 branch history linear with no merges;
• After your approval, my changes should be squashed into one CL and merged into the main repo's main branch;
• I will periodically sync the main repo main branch to my own main branch and rebase my feature branch to it;

LMK if this works well for you reviewers. My goal is to keep a simple workflow and help keep the main repo main branch history linear (ideally with no/few merges, but only rebases).

@vipannalla @wyzhang ...

[xy12181]

looks good

Thank you @vipannalla for your quick turnaround!

[xy12181]

@khatwanimohit, @gobbleturk, @RissyRan, @khatwanimohit could any of you please review and merge this PR? thank you!