Feat: Correct context length for models

[hahuyhoang411]

Describe Your Changes

• Correct upper bound for models on the Catalog

Fixes Issues

• Closes #
• Closes #

Self Checklist

• Codeninja 4k -> 8k
• CmdR 4k -> 130k
• Package bump

[hahuyhoang411]

I think we need a dynamic sync between upper bound of max_len and ctx_len.

• My approach right now is to set max_len = ctx_len in the model.json file to display the maximum but we need to handle this dynamically

• The upper bound of max_len should also be the same as configured ctx_len

• E.g. ctx_len = 2048, max_len should be 2048

• Otherwise, it would confuse the user like this:

[hahuyhoang411]

Thanks @urmauur for the fast work

#2870

[Van-QA]

LGTM