feat: Switch to TRT-LLM LLM (High Level) API over trtllm-build CLI workflow

[rmccorm4]

Changes:

• Switch to TRT-LLM LLM (High Level) API over trt-llm build workflow
  • Remove unnecessary helper scripts
  • Expands model support
  • Simplifies engine generation and model download process
• Refer to framework model support sections in README
• Add some better checks and error messages around --source arg
• Add 'local:' source prefix for parity
• Bump version to 0.1.0dev so it's easier to have 'patch' versions such as 0.1.1, 0.1.2, etc.
• Disable genai-perf (triton profile) based tests due to bug in genai-perf r24.09:

Note:

• Disable genai-perf (triton profile) based tests due to bug in genai-perf r24.09:

$ triton profile -m gpt2
triton - INFO - Running: 'genai-perf profile -m gpt2'
2024-10-25 12:40 [INFO] genai_perf.parser:83 - Profiling these models: gpt2
Traceback (most recent call last):
  File "/home/rmccormick/.local/lib/python3.10/site-packages/genai_perf/main.py", line 198, in run
    generate_inputs(config_options)
  File "/home/rmccormick/.local/lib/python3.10/site-packages/genai_perf/main.py", line 125, in generate_inputs
    inputs.create_inputs()
  File "/home/rmccormick/.local/lib/python3.10/site-packages/genai_perf/inputs/inputs.py", line 57, in create_inputs
    json_in_pa_format = self._convert_generic_json_to_output_format(
  File "/home/rmccormick/.local/lib/python3.10/site-packages/genai_perf/inputs/inputs.py", line 77, in _convert_generic_json_to_output_format
    return converter.convert(generic_dataset, self.config)
  File "/home/rmccormick/.local/lib/python3.10/site-packages/genai_perf/inputs/converters/tensorrtllm_converter.py", line 52, in convert
    pa_json = self._populate_trtllm_output_json(
  File "/home/rmccormick/.local/lib/python3.10/site-packages/genai_perf/inputs/converters/tensorrtllm_converter.py", line 96, in _populate_trtllm_output_json
    pa_json = self._add_optional_tags_to_trtllm_json(
  File "/home/rmccormick/.local/lib/python3.10/site-packages/genai_perf/inputs/converters/tensorrtllm_converter.py", line 125, in _add_optional_tags_to_trtllm_json
    if config.add_stream:
AttributeError: 'str' object has no attribute 'add_stream'

[KrishnanPrash]

Overall change LGTM. Will clarify order of operations and upcoming releases offline.