Add table-transformer to NormalizedConfigManager

[kratos0718]

Summary

Part of #351. table-transformer already has an ONNX export config but was missing from NormalizedConfigManager, so ORTOptimizer can't run on it yet.

• Added "table-transformer": NormalizedTextConfig to _conf (alphabetically, between splinter and t5)
• Removed table-transformer from the TODO: missing normalized configs comment above it

Same situation as DETR (#2510, also open on this issue): table-transformer is DETR-based, so its config stores transformer stats under d_model/encoder_attention_heads/encoder_layers, but TableTransformerConfig.attribute_map already aliases these to hidden_size/num_attention_heads/num_hidden_layers, so plain NormalizedTextConfig resolves them with no custom with_args(...) mapping needed.

Test plan

• Added two tests to tests/utils/test_normalized_config.py: one confirming table-transformer resolves to NormalizedTextConfig, and one instantiating a real TableTransformerConfig() and asserting the normalized hidden_size/num_attention_heads/num_layers match d_model/encoder_attention_heads/encoder_layers
• Ran both locally against the actual transformers.TableTransformerConfig (not mocked) — both pass