fix: implement lazy client creation in replicate.use()

[zeke]

Problem

replicate.use() was creating a client immediately, causing errors when no API token was available at call time. This prevented usage in Cog pipelines where tokens are provided through cog.current_scope() at runtime.

Solution

Modified Function and AsyncFunction classes to support lazy client creation by always accepting client classes and instantiating them on demand.

Changes

• _module_client.py: Pass Replicate and AsyncReplicate classes directly instead of factory functions
• lib/_predictions_use.py: Simplified to always use Type[Client] approach:
  • Removed Union[Client, Type[Client]] in favor of just Type[Client]
  • Updated constructors to store _client_class instead of mixed types
  • Simplified _client property to always instantiate from class
  • Updated all use() overloads to accept class types only
  • Use issubclass() for async client detection
• Added simple test demonstrating the fix

Result

✅ replicate.use("model") works even when no token is initially available
✅ Client created lazily when model is actually called
✅ Token retrieved from current cog scope at call time
✅ Simplified, consistent approach using only class types
✅ Clean type inference without complex Union handling

Fixes the original error:

replicate.ReplicateError: The bearer_token client option must be set either by passing bearer_token to the client or by setting the REPLICATE_API_TOKEN environment variable

[aron]

Looks fine, but I think we can simplify it a little, I've left some suggestions inline.

[aron]

Looks much cleaner, thanks. Left some comments about the unit tests, consider them suggestions.

[aron]

Nice, looks good.

[aron]

Minor tweak, we want to assert that the same model refreshes the client between calls. If we're looking at the client we can look at the bearer_token to verify it has actually picked up the token correctly.

Suggested change

	# Access the client property - this should trigger client creation and cog.current_scope call
	_ = model._client
	assert mock_cog.current_scope.call_count == 1
	# Change the token and access client again - should trigger another call
	mock_context.items.return_value = [("REPLICATE_API_TOKEN", "test-token-2")]
	# Create a new model to trigger another client creation
	model2 = replicate.use("test/model2") # type: ignore[misc]
	_ = model2._client
	# Access the client property - this should trigger client creation and cog.current_scope call
	assert model._client.bearer_token == "test-token-1"
	assert mock_cog.current_scope.call_count == 1
	# Change the token and access client again - should trigger another call
	mock_context.items.return_value = [("REPLICATE_API_TOKEN", "test-token-2")]
	# Assert that the second time picks up the new token
	assert model2._client.bearer_token == "test-token-2"