test: upgrade test app to net10 and split benchmarks into a dedicated project

fix: harden BPE loading with cache fallback and in-memory fallback

Update README.md

update action

Optimize memory allocation and execution efficiency.

Add Support gpt-5 gpt-4.5 models.

Modified tikToken.Encode in Program.cs to include allowedSpecialMode for special token handling.

Update action.

Optimize model name matching encoding.

Merge branch 'master' of https://github.com/aiqinxuancai/TiktokenSharp

Add support o1 models.

Update README.md

Update readme

Update description

Fix match model to encoding error.