Fixes delayed sampling for sequential requests

[attafosu]

• Previously LLM.generate() could not be called multiple times with delayed sampling enabled.
• This also was the case with step() calls
• Issue occurs when after the last (batch) request is finished, and we're starting a new request, but cached_step_inputs and cached_step_outputs still contain elements saved from the last served (batch) request. This shouldn't be the case.
• The cleanest solution would be to skip appending to cached_step_inputs/outputs if the recently generated output is the final token generated for the current batch request. But couldn't find a cleaner way to check for this in the model runner.
• So we instead check (in _patch_prev_output) for when the scheduler context has empty output_queue, which means no pending outputs to patch.

Tests here: https://github.com/habana-internal/mlperf_inference/pull/158