Gracefully handle legacy Sequential configs

[GabrielBianconi]

Summary

In a recent change to Keras (#11133) incorporated in 2.2.3, name was added to the Sequential config. From the release notes:

Breaking Changes
Modify the return value of Sequential.get_config(). Previously, the return value was a list of the config dictionaries of the layers of the model. Now, the return value is a dictionary with keys layers, name, and an optional key build_input_shape. The old config is equivalent to new_config['layers']. This makes the output of get_config consistent across all model classes.

However, Keras now raises an UnboundLocalError exception whenever loading a legacy configuration file. In from_config, name can be undefined if this variable is not present in the config, and in line 298 of the modified file, is used by model = cls(name=name).

I modified the code such that legacy configs still load, but throws a warning. Maybe a different behavior is desired, but Keras, in my opinion, shouldn't crash like this (especially without any warnings).

Related Issues

#11133

PR Overview

• This PR requires new unit tests [y/n] (make sure tests are included)
• This PR requires to update the documentation [y/n] (make sure the docs are up-to-date)
• This PR is backwards compatible [y/n]
• This PR changes the current API [y/n] (all API changes need to be approved by fchollet)

[fchollet]

Sounds good, thanks.

[gabrieldemarmiesse]

Maybe we should have migration tests to ensure that we don't break the loading of saved models. I am unsure how to implement this though.