Add PPR relation token features to graph transformer

[mkolodner-sc]

Summary

Adds Graph Transformer support for generic PPR edge-attribute metadata.

PPR can emit multi-column edge_attr tensors where column 0 is the scalar PPR weight and the remaining columns contain sampler metadata. This PR lets Graph Transformer consume those extra columns as token-input features while preserving the existing scalar PPR path.

What This Enables

A PPR edge can now carry:

• ppr_weight: the first edge_attr column, used as the scalar PPR score.
• ppr_features: all remaining edge_attr columns, projected into the token embedding as continuous token features.

This allows PPR metadata such as hop distance, channel scores, channel min-hop values, or channel presence bits to flow into Graph Transformer tokens without changing how scalar PPR batches behave.

Changes

• Adds reserved feature name ppr_features.
• Supports scalar and multi-column PPR edge_attr.
• Keeps ppr_weight mapped to edge_attr[..., 0].
• Maps ppr_features to edge_attr[..., 1:].
• Preserves raw metadata values in ppr_features, including hop values greater than 1.
• Keeps ppr_weight clamped as a scalar score in [0, 1].
• Sorts PPR tokens by descending edge_attr[..., 0] and carries feature columns through the same permutation.
• Validates that ppr_features is used only as token input, not attention bias.
• Updates token-input handling to support multi-column continuous features.
• Adds transform and encoder tests for PPR feature extraction, ordering, and forward-pass support.

Toy Example

Suppose typed PPR emits PPR edges from an anchor to selected tokens.

Each PPR edge represents one selected (anchor, token) pair. The edge_attr row describes how PPR reached/scored that token:

• Column 0 is the scalar PPR weight.
• Column 1 is the global minimum hop from the anchor.
• The next block contains per-channel PPR scores.
• The next block contains per-channel minimum hops.
• The final block contains per-channel presence bits.

For example, with two typed PPR channels:

data["user", "ppr", "item"].edge_attr = torch.tensor([
    # best_ppr, global_min_hop, ch0_score, ch1_score, ch0_min_hop, ch1_min_hop, ch0_present, ch1_present
    [0.90,     1.0,            0.90,      0.00,      1.0,         0.0,         1.0,         0.0],
    [0.75,     2.0,            0.10,      0.75,      2.0,         2.0,         1.0,         1.0],
    [0.40,     4.0,            0.00,      0.40,      0.0,         4.0,         0.0,         1.0],
])

Graph Transformer interprets this as:

ppr_weight = edge_attr[:, 0:1]
ppr_features = edge_attr[:, 1:]

A model can consume both as token-input features:

encoder = GraphTransformerEncoder(
    ...,
    sequence_construction_method="ppr",
    anchor_based_input_attr_names=[
        "ppr_weight",
        "ppr_features",
    ],
)

These continuous features are concatenated, passed through the learned token-input projection, and added to the corresponding token embeddings before the Transformer layers run.

Validation

• Added focused unit coverage for multi-column PPR edge attrs and Graph Transformer token-input consumption.
• Verified stale ppr_relation_features references were removed.
• Verified ppr_features preserves raw metadata values rather than clamping hop-like columns.
• Ran syntax and whitespace checks on the changed Graph Transformer files and tests.

[yliu2-sc]

isn't ppr_reserved_feature_names always true since you set it above? Should it only be set when PPR mode?

[yliu2-sc]

same