🐛 [Bug] require_full_compilation never reaches partitioner

[dgcnz]

I'm not sure if this is intended, but it seemed odd to me that there is logic inside both fast and global partitioners (see snippets 1 and 2) with require_full_compilation but the compile_module never feeds the passes the flag as parameter (see snippet 3).

Snippets

TensorRT/py/torch_tensorrt/dynamo/partitioning/_adjacency_partitioner.py

Lines 206 to 209 in fa02fd3

	if not full_support and self.require_full_compilation:
	raise AssertionError(
	"require_full_compilation=True was specified, but model is not fully supported"
	)

TensorRT/py/torch_tensorrt/dynamo/partitioning/_global_partitioner.py

Lines 68 to 71 in fa02fd3

	if not full_support and self.require_full_compilation:
	raise AssertionError(
	"require_full_compilation=True was specified, but model is not fully supported"
	)

TensorRT/py/torch_tensorrt/dynamo/_compiler.py

Lines 371 to 388 in fa02fd3

	if settings.use_fast_partitioner:
	try:
	logger.info("Partitioning the graph via the fast partitioner")
	partitioned_module, supported_ops = partitioning.fast_partition(
	gm,
	verbose=settings.debug,
	min_block_size=settings.min_block_size,
	torch_executed_ops=settings.torch_executed_ops,
	)
	except torch.fx.passes.splitter_base.FxNetSplitterInternalError:
	logger.error(
	"Partitioning failed on the subgraph with fast partition. See trace above. "
	"Retrying with global partition.",
	exc_info=True,
	)
	fast_partitioner_failed = True
	settings.use_fast_partitioner = False