Make logit_bias work for all gpt-based multimodal models & Add supports of gpt-4o tokenizer for reverse_proxy (#418)

# PR Checklist
- [ ] Did you check if it works normally in all models? *ignore this
when it dosen't uses models*
- [ ] Did you check if it works normally in all of web, local and node
hosted versions? if it dosen't, did you blocked it in those versions?
- [ ] Did you added a type def?

# Description
I fix the code which now make exception for gpt-based models so it can
send logit_bias to server and gpt4o with reverse_proxy automatically
choose new tokenizer.

src/ts/process/request.ts

@@ -495,12 +495,12 @@ export async function requestChatDataMain(arg:requestDataArgument, model:'model'
 
             if(supportsInlayImage()){
                 // inlay models doesn't support logit_bias
-                // gpt-4-turbo supports both logit_bias and inlay image
+                // OpenAI's gpt based llm model supports both logit_bias and inlay image
                 if(!(
-                    aiModel.startsWith('gpt4_turbo') || 
+                    aiModel.startsWith('gpt') || 
                     (aiModel == 'reverse_proxy' && (
-                        db.proxyRequestModel?.startsWith('gpt4_turbo') ||
-                        (db.proxyRequestModel === 'custom' && db.customProxyRequestModel.startsWith('gpt-4-turbo'))
+                        db.proxyRequestModel?.startsWith('gpt') ||
+                        (db.proxyRequestModel === 'custom' && db.customProxyRequestModel.startsWith('gpt'))
                     )))){
                     // @ts-ignore
                     delete body.logit_bias

src/ts/tokenizer.ts

@@ -35,6 +35,11 @@ export async function encode(data:string):Promise<(number[]|Uint32Array|Int32Arr
             case 'llama3':
                 return await tokenizeWebTokenizers(data, 'llama')
             default:
+                // Add exception for gpt-4o tokenizers on reverse_proxy
+                if(db.proxyRequestModel?.startsWith('gpt4o') || 
+                (db.proxyRequestModel === 'custom' && db.customProxyRequestModel.startsWith('gpt-4o'))) {
+                    return await tikJS(data, 'o200k_base')
+                }
                 return await tikJS(data)
         }
     }