[Conformal EEG] Add TFM-Tokenizer model for EEG signal tokenization

[lehendo]

Contributor: Arjun Chatterjee (arjunc4@illinois.edu)
Contribution Type: Model

Description
Merge Jathurshan's model TFM-Tokenizer to PyHealth and be able to load his pre-trained weights

Files to review
pyhealth/models/tfm_tokenizer.py - New TFM-Tokenizer model implementation including:

TFMTokenizer - Main PyHealth-integrated model class
TFM_VQVAE2_deep - Tokenizer with frequency and temporal encoding
TFM_TOKEN_Classifier - Token-based classifier for downstream tasks
EMAVectorQuantizer - Exponential moving average VQ-VAE quantizer
Helper methods: get_embeddings(), get_tokens(), load_pretrained_weights()

pyhealth/models/init.py - Added imports for TFMTokenizer and related classes
pyproject.toml - Added dependencies: einops>=0.8.0, linear-attention-transformer>=0.19.1

[Jathurshan0330]

The code is good and replicates tfmtokenizer well