Support custom imported module serialization with cloudpickle

[erandeutsch]

Description

This pull request implements the ability to pickle imported modules by value when saving whole programs.

This aims to solve issue #8285; a saved dspy program cannot be reliably loaded from a different directory if it uses an imported signature from another file.

Changes

A new optional modules_to_serialize parameter is added to the save method, allowing users to specify custom imported modules for serialization with cloudpickle. modules_to_serialize is a list of modules, which are then registered for serialization by value using cloudpickle.register_pickle_by_value.

For example, saving a module which uses an imported module:

module.save("testModule", save_program=True, modules_to_serialize=[ExtractInfo])

Using this, modules can be loaded using dspy.load in a separate directory without any problems related to importing modules. This improves the ability to share dspy programs across computers and repositories, without any import issues.

[chenmoneygithub]

Thanks for the PR! Left 2 minor comments, otherwise LGTM!

[erandeutsch]

I added a section to the "Saving and Loading your DSPy program" tutorial explaining how to use the new modules_to_serialize parameter in the save method for serializing imported modules.

dspy/docs/docs/tutorials/saving/index.md

Lines 99 to 129 in e4b7148

	### Serializing Imported Modules
	When saving a program with `save_program=True`, you might need to include custom modules that your program depends on.
	You can specify which custom modules should be serialized with your program by passing them to the `modules_to_serialize`
	parameter when calling `save`. This ensures that any dependencies your program relies on are included during serialization and
	available when loading the program later.
	This uses cloudpickle's `cloudpickle.register_pickle_by_value` function in order to register a module as picklable by value. When
	a module is registered this way, cloudpickle will serialize the module by value rather than by reference, ensuring that the
	module contents are preserved with the saved program.
	For example, if your program uses custom modules:
	```python
	import dspy
	import my_custom_module
	compiled_dspy_program = dspy.ChainOfThought(my_custom_module.custom_signature)
	# Save the program with the custom module
	compiled_dspy_program.save(
	"./dspy_program/",
	save_program=True,
	modules_to_serialize=[my_custom_module]
	)
	```
	This ensures that the required modules are properly serialized and available when loading the program later. Any number of
	modules can be passed to `modules_to_serialize`. If you don't specify `modules_to_serialize`, no additional modules will be
	registered for serialization.

[chenmoneygithub]

LGTM!