Ensure essential packages are not detached!

[zhongzheng1999]

In this branch, I removed the part detaching packages in scib.pp.normalize because it would lead to unpredictable errors when using rpy2 to run R code.
After learning from my own experience, I want to advise against detaching R packages casually. Instead, after deleting rm(list=ls()), running gc() usually suffices.
A simple example: in the final part of scib.pp.normalize, detaching most R packages, including scran and its dependency package S4Vector, would result in errors such as "cannot coerce class 'integer' to a DataFrame" if a loop code block is set up. Even if you reload scran (which depends on S4Vector) or S4Vector, the error persists and can only be resolved by restarting your kernel. Next, I'll demonstrate this issue with a simple example.

example.md

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 54.66%. Comparing base (9271d92) to head (afbaf42).
Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #402      +/-   ##
==========================================
+ Coverage   54.60%   54.66%   +0.06%     
==========================================
  Files          22       22              
  Lines        1683     1681       -2     
==========================================
  Hits          919      919              
+ Misses        764      762       -2     

Flag	Coverage Δ
unittests	54.66% <ø> (+0.06%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
scib/preprocessing.py	19.06% <ø> (+0.10%)	⬆️