feat(langchain): improve instrumentation of input messages

[shellmayr]

• Improve the instrumentation of input messages in the LangChain SDK

Before:

After:

Closes TET-1058

[codecov]

Codecov Report

❌ Patch coverage is 48.97959% with 25 lines in your changes missing coverage. Please review.
✅ Project coverage is 84.77%. Comparing base (f702ec9) to head (c3eada8).
⚠️ Report is 5 commits behind head on master.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
sentry_sdk/integrations/langchain.py	48.97%	18 Missing and 7 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4750      +/-   ##
==========================================
- Coverage   84.89%   84.77%   -0.12%     
==========================================
  Files         156      157       +1     
  Lines       16115    16179      +64     
  Branches     2741     2757      +16     
==========================================
+ Hits        13681    13716      +35     
- Misses       1645     1668      +23     
- Partials      789      795       +6     

Files with missing lines	Coverage Δ
sentry_sdk/integrations/langchain.py	63.99% <48.97%> (-1.69%)	⬇️

... and 7 files with indirect coverage changes

[linear]

TET-1058 Fix JSON rendering

[cursor]

Bug: Double Serialization in set_data_normalized

The set_data_normalized function at line 735 receives data that's already safe_serialized. Since set_data_normalized handles its own serialization, this can cause double serialization or incorrect data. This is inconsistent with other safe_serialize changes in the commit.