Feat/implement streaming path in audio transcription (#926)

* implement streaming path in audio transcription endpoint

Signed-off-by: WaelRabah11 <wael.rabah@multiversecomputing.com>

* Refactored proxy_multipart_request to share post call, headers, and error
  handling between streaming and non-streaming paths. Stats hooks
  (on_new_request/on_request_response/on_request_complete) now fire on both paths.
  Tutorial fixed to use with open(...). Tests pass (8/8).

Signed-off-by: WaelRabah11 <wael.rabah@multiversecomputing.com>

* Update src/vllm_router/services/request_service/request.py

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Signed-off-by: Wael Rabah <47258198+WaelRabah@users.noreply.github.com>

* update streaming example, refactor try/except to reduce nesting and catch json decode errors

Signed-off-by: WaelRabah11 <wael.rabah@multiversecomputing.com>

* Added backend_response.close() to the non-streaming finally block

Signed-off-by: WaelRabah11 <wael.rabah@multiversecomputing.com>

* fix outdated test

Signed-off-by: WaelRabah11 <wael.rabah@multiversecomputing.com>

---------

Signed-off-by: WaelRabah11 <wael.rabah@multiversecomputing.com>
Signed-off-by: Wael Rabah <47258198+WaelRabah@users.noreply.github.com>
Co-authored-by: WaelRabah11 <wael.rabah@multiversecomputing.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>