Add trace with JIT example

[mawad-amd]

Initial version of trace integration with JIT. Here is the output:

In code, it looks like this:

mlir-aie/programming_examples/basic/vector_scalar_mul/vector_scalar_mul_jit.py

Lines 179 to 190 in 612af49

	if args.trace_size > 0:
	trace.set_trace_size(args.trace_size)
	trace.start_trace()
	# JIT-compile the kernel then launches the kernel with the given arguments
	vector_scalar_mul(input0, scalar, output)
	# Stop tracing and save results if tracing was enabled
	if args.trace_size > 0:
	trace_filename = f"trace_output_{args.num_elements}_{args.data_type}.json"
	trace.stop_trace(trace_filename)
	print(f"Tracing completed and saved to {trace_filename}")

And this:

mlir-aie/programming_examples/basic/vector_scalar_mul/vector_scalar_mul_jit.py

Lines 115 to 126 in 612af49

	worker = Worker(
	core_body,
	fn_args=[of_in.cons(), of_factor.cons(), of_out.prod(), scale],
	trace=1 if trace.get_trace_size() > 0 else 0,
	)
	# Runtime operations to move data to/from the AIE-array
	rt = Runtime()
	with rt.sequence(tensor_ty, scalar_ty, tensor_ty) as (A, F, C):
	if trace.get_trace_size() > 0:
	rt.enable_trace(trace.get_trace_size())

I would have liked to hide this away too but it's not straight forward at the moment because of how the default workflow works (i.e., I can't simply call enable_trace for the runtime sequence if tracing is enabled in the backend). There are also other complications for tracing that Jack has brought up.

Output trace visualization:

[mawad-amd]

@jackl-xilinx @AndraBisca @jgmelber This is an initial draft for trace support in JIT.