Molgraph

Minor features and improvements

- `molgraph.models.interpretability`
- `GradientActivationMapping` deprecates `layer_names` and will be default watch the node features of all graph tensors.

Bug fixes
- `molgraph.tensors`
- `GraphTensor` can now add field with prepended underscore in the input pipeline (after batching etc.)

Bug fixes

- `molgraph.models`
- `molgraph.models.interpretability` models now work with multi-label data, as it was supposed to.
- `molgraph.models.interpretability.GradientActivationMapping` now computes alpha correctly, namely, computed for each subgraph separately (based on the graph indicator).

Minor features and improvements

- `molgraph.models`
- `molgraph.models.interpretability` models are now simplified and are not by default wrapped in tf.function; if desirable, the user may wrap it in tf.function themselves.

Bug fixes

- `molgraph.layers`
- `molgraph.layers.GNNLayer`'s get_config and from_config methods are updated to allow for serialization of GNN models.
- `molgraph.layers.GNNInputLayer` and `molgraph.layers.GNNInput` were added to allow for serialization of GNN models.
- `molgraph.layers.StandardScaling`, `molgraph.layers.Threshold` and `molgraph.layers.CenterScaling` can now be loaded.
- `molgraph.chemistry`
- `molgraph.chemistry.Tokenizer` now appropriately adds self loops (if specified).

Fixes
- `molgraph.layers.GATv2Conv` should now better correspond to the original GATv2 implementation.

Bug fixes
- MolGraph should now install appropriate tensorflow version.