[Docs] Add Relax VM architecture documentation

[tlopex]

• Add a dedicated architecture document (docs/arch/relax_vm.rst) for the Relax Virtual
  Machine, covering the compilation pipeline (Relax IR → VMCodeGen → VMExecutable), the
  4-opcode instruction set and encoding format, the register-based execution model (VMFrame,
  RunLoop, VMClosure dispatch), built-in operations, serialization, and the Python-level
  interface (direct invocation, stateful API, profiling, instrumentation).
• Refactor the inline VM section in docs/arch/index.rst into a brief summary with a
  cross-reference to the new page, and add relax_vm to the toctree.

[gemini-code-assist]

Code Review

This pull request introduces a comprehensive architecture document for the Relax Virtual Machine (VM) and updates the main architecture index to point to this new resource. The new documentation covers the end-to-end compilation flow, instruction set, executable format, and runtime execution model. Feedback provided focuses on improving the accuracy of the binary format description and ensuring code snippets are complete by adding missing imports.

[gemini-code-assist]

The documentation mentions four sections in the binary format, but according to include/tvm/runtime/vm/executable.h (lines 183-203), there are five sections: globals, memory scopes, constant pool, bytecode, and packed function names. It would be more accurate to include the packed function names section in this list.

Suggested change

	The binary format includes a magic number (``0xD225DE2F4214151E``), a version string
	(currently ``"0.14"``), followed by four sections: globals (the function table), memory scopes,
	constant pool, and bytecode. ``AsText()`` and ``AsPython()`` provide human-readable representations
	The binary format includes a magic number (``0xD225DE2F4214151E``), a version string
	(currently ``"0.14"``), followed by five sections: globals (the function table), memory scopes,
	constant pool, bytecode, and packed function names. ``AsText()`` and ``AsPython()`` provide

[gemini-code-assist]

The instrumentation example uses VMInstrumentReturnKind.NO_OP, but VMInstrumentReturnKind is not imported in the snippet. For completeness and to make the example directly runnable, please add the necessary import from tvm.runtime.vm.

Suggested change

	if before_run:
	print(f"About to call: {func_symbol}")
	return VMInstrumentReturnKind.NO_OP
	vm.set_instrument(my_instrument)
	vm["main"](inp)
	from tvm.runtime.vm import VMInstrumentReturnKind
	def my_instrument(func, func_symbol, before_run, ret_value, *args):
	if before_run:
	print(f"About to call: {func_symbol}")
	return VMInstrumentReturnKind.NO_OP
	vm.set_instrument(my_instrument)
	vm["main"](inp)

[tlopex]

cc @MasterJH5574