Fix use of LoRA tokenizers

[njhill]

This depends on vLLM PR vllm-project/vllm#6227, which is only held up because I need to find time to add a test to it.

[dtrifiro]

Merge after #47

[codecov-commenter]

Codecov Report

❌ Patch coverage is 33.33333% with 34 lines in your changes missing coverage. Please review.
✅ Project coverage is 30.75%. Comparing base (d32cfb1) to head (8aa1f95).
⚠️ Report is 253 commits behind head on main.

Files with missing lines	Patch %	Lines
src/vllm_tgis_adapter/grpc/grpc_server.py	14.28%	23 Missing and 1 partial ⚠️
tests/conftest.py	60.00%	7 Missing and 1 partial ⚠️
tests/test_grpc_server.py	33.33%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main      #40       +/-   ##
===========================================
- Coverage   62.96%   30.75%   -32.21%     
===========================================
  Files          18       18               
  Lines        1250     1278       +28     
  Branches      222      228        +6     
===========================================
- Hits          787      393      -394     
- Misses        390      881      +491     
+ Partials       73        4       -69     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.