Manage the vLLM lifecycle

[tobrun]

This is a first step towards improving the usability and robustness of this repository. This change ensures we clean up all the vLLM instances when we exit:

• finished executing the task
• error occurs
• keyboard interruption

imo, this isn't the optimal solution. I personally would love to divide the LLM engine lifetime from the actual job processing but that is for a future PR.

closes #109