Update on "[ExecuTorch][Weight Sharing] Track Named Data Store in EdgeProgramManager"

We enable Backends to return Named Data by adding NamedDataStoreOutput to the preprocess result. This is a completely BC change, as no backends with an implemented preprocess will see any change if nothing is explicitly implemented.

For backend developers to leverage the new NamedDataStore, they can initialize a new NamedDataStore() within preprocess, add_named_data to the data store, and return the NamedDataStore.get_named_data_store_output() in the preprocess result like such:

```
def preprocess(ExportedProgram, List[CompileSpecs]) -> PreprocessResult:
    named_data_store = NamedDataStore()

    for node in exported_program.graph.nodes:
        named_data_store.add_named_data("name", bytes)

    return PreprocessResult(
        processed_bytes=bytes,
        debug_handle_map={},
        data_store_output= named_data_store.get_named_data_store_output()
    )
```


Under the hood, the data store output is embedded in the loweredbackendmodule, (serializing loweredbackendmodule by itself with the a named_data_store_output is still a todo). But via the EdgeProgramManager path, we add the named_data_store_outputs to the edge_program_manger's named data store to keep track of all the named data returned by backends.

Differential Revision: [D70451660](https://our.internmc.facebook.com/intern/diff/D70451660/)

[ghstack-poisoned]

Author: mcr229

exir/backend/test/test_backend_with_named_data_map.py

@@ -67,7 +67,7 @@ def false_branch(self, x):
 
             def forward(self, x, y):
                 z = x / y
-                z = torch.cond(z > 1, self.true_branch, self.false_branch, [x])
+                z = torch.cond(z.sum() > 0, self.true_branch, self.false_branch, [x])
                 return z - z
 
         ep = to_edge(torch.export.export(M(), (torch.randn(1, 2), torch.randn(1, 2))))