Undeprecate literal_(de)stringizer

[rossbar]

I propose to rescind the deprecation of gml.literal_stringizer and gml.literal_destringizer.

These functions have been deprecated since 2.5 but were never removed because it is not possible to do so without breaking tests. I spent a frustrating couple hours diving into this and gained a few insights. Currently, literal_stringizer is necessary to handle two specific cases that are tested:

1. Graphs with nodes that are tuples, and
2. Attributes that include nested sequences

The first one is relatively easy to "fix" without literal stringizer. For completeness sake, it would involved adding the following snippet:

elif isinstance(value, tuple) and key == "label":
    yield indent + key + f" \"({','.join(repr(v) for v in value)})\""

before L761.

The 2nd one is really tough to crack. The default stringizer function has an in_list function to decide how to handle elements inside lists, but this only works for "1D" lists - nested sequences break this pattern and will end up raising an exception. literal_stringizer gets around this by only unpacking the outermost list, then directly repr-ing everything inside it. I'm not even sure that the GML standard supports nested lists as attributes! However, this currently "works" according to test_data_types, even if the result isn't actually valid, standards-compliant gml.

IMO, the simplest thing to do would be to revert the deprecation. On the off-chance that there are users depending on this behavior, literal_stringizer is currently the only simple way to achieve it. It looks like the original motivation for deprecating the literal_*stringizer functions was Python 2->3 cleanup, but these corner cases are still relevant and unrelated to Python 2. Furthermore, I think reverting the deprecation will be relatively painless, as the deprecation warnings aren't currently visible due to the stack level.

[dschult]

Thanks Ross!
I appreciate the effort spent diving into these functions. :)

[jarrodmillman]

Thanks!! I spent too much time to trying to sort this out as well.