BUG: The bytes returned by cuda StreamReader may leads to deserialization failed

[ChengjieLi28]

What do these changes do?

When I adapt copy_to to xorbits, xoscar cannot deserialize the transferred cupy object.
In xorbits, all the cuda object will use rmm.DeviceBuffer-based object to store actual data.
But in xoscar, the result of deserialization could be the organic bytes object.
I think using rmm.DeviceBuffer is definitely correct.

Check code requirements

• tests added / passed (if needed)
• Ensure all linting tests pass

[codecov]

Codecov Report

Merging #40 (a23008e) into main (7531e22) will decrease coverage by 0.15%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##             main      #40      +/-   ##
==========================================
- Coverage   93.79%   93.64%   -0.15%     
==========================================
  Files          43       43              
  Lines        3399     3399              
  Branches      675      675              
==========================================
- Hits         3188     3183       -5     
- Misses        139      145       +6     
+ Partials       72       71       -1     

Flag	Coverage Δ
unittests	93.49% <ø> (-0.15%)	⬇️

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

Impacted Files	Coverage Δ
python/xoscar/backends/communication/utils.py	100.00% <ø> (ø)

... and 2 files with indirect coverage changes

[aresnow1]

LGTM

[codingl2k1]

LGTM