chore: summarization logic regarding Anthropic

[Illusion2-1]

Users using the Claude model provided by Anthropic may find that they are using the same summary model as the dialogue model, which generally results in additional costs.

This PR adds summary model judgement logic that was missing when using the Claude model, and by default the user's dialogue will be summarised using the haiku, which is the least costly one.

Of course, if the user wishes to use a different summarisation model, they can change it themselves.
ANTHROPIC_SUMMARIZE_MODEL.

[vercel]

@Illusion2-1 is attempting to deploy a commit to the NextChat Team on Vercel.

A member of the Team first needs to authorize it.

[github-actions]

Your build has completed!

Preview deployment

[Dean-YZG]

currentlly,only internal Claude Models supported,so just identifying claude models by prefix 'claude' is not right

[Illusion2-1]

Appreciation for the review! However, I was a little confused; Are 'internal Claude models' these?

const anthropicModels = [
  "claude-instant-1.2",
  "claude-2.0",
  "claude-2.1",
  "claude-3-sonnet-20240229",
  "claude-3-opus-20240229",
  "claude-3-haiku-20240307",
];

In that case, I didn't see identifying via prefix has any problem...
If that means custom models (some users may be using reverse proxy for claude but with OpenAI API) should be supported as well, will using includes or regex be appropriate?

includes

if (currentModel.toLowerCase().includes("claude") || currentModel.toLowerCase().includes("anthropic")){
    return ANTHROPIC_SUMMARIZE_MODEL;
}

regex

if (/claude|anthropic/i.test(currentModel)) {
    return ANTHROPIC_SUMMARIZE_MODEL;
}