Eagle3 fallback attention mask crashes with KV cache

## Problem

`prepare_4d_causal_attention_mask` expects the padding-mask key dimension to match `kv_len`. In the Eagle3 TTT loop with a KV cache, `kv_len` grows with cached chunks, but the incoming 2D `attention_mask` still covers only the current chunk.

For `seq_len = 4` at cached step `k = 1`:

```text
causal mask: [1, 1, 4, 8]
expanded mask: [B, 1, 1, 4]
```

Adding these tensors fails on the key dimension.

## Impact

The non-flex fallback attention path crashes whenever cached Eagle3 TTT steps require `kv_len > seq_len`.

## Fix

Fixed in PR #30 by extending the 2D keep-mask to `kv_len` before converting it to the additive 4D mask. The final fix repeats the fixed chunk mask pattern, preserving padded cached columns instead of marking all cached positions valid.