🐛 Bug description (make it clear to everyone)

The code uses rstrip("api") to drop a trailing "/api" from a base URL/host.
But rstrip treats its argument as a set of characters ({'a','p','i'}) and strips all trailing occurrences of those chars.
This mutilates valid hosts/paths.
Example:

"www.example.ai".rstrip("api")  # -> "www.example."

🧑‍💻 Steps to reproduce

1. Set a host/base URL ending with .ai (or anything ending in a, p, or i), e.g.:

   base = "https://www.example.ai/api"

2. Run the logic that normalizes/removes the "/api" suffix (e.g., login/init flow).
3. Observe the resulting host string becomes:

   https://www.example.
   

   (the trailing ai is incorrectly stripped)

👾 Expected result

• Remove the suffix only if the string ends with the exact "/api" once.
• Do not strip characters from legitimate domain endings such as .ai, .app, etc.

Examples:

• "https://www.example.ai/api" → "https://www.example.ai"
• "https://www.example.ai" → unchanged

🚑 Any additional info (screenshots, code, logs)

Minimal proof:

>>> "www.example.ai".rstrip("api")
'www.example.'
>>> "https://www.example.ai/api".rstrip("api")
'https://www.example.'

🔧 Suggested fix

🧭 Environment

• SwanLab Version: v0.6.7
• Platform/OS: Ubuntu 22.04
• Python Version: 3.10.12

Reported by Kingdee Cloud Team 💼